**** BEGIN LOGGING AT Mon Feb 04 02:59:59 2013 Feb 04 03:41:29 erm, do we have something like "depends on A *or* B"? Feb 04 03:42:25 bnf would now depend on *either* ic2-tools *and/or* feature bq27x00 module Feb 04 03:42:40 no idea how to set such dependency, though Feb 04 03:43:44 of course I could bump ic2-tool to provide some feature like bq27x00-chip-info, and ask Pali to make kernel-power provide the same, then, make bnf depend on it... Feb 04 03:43:49 still, it sounds silly Feb 04 03:49:51 I've only ever seen optional dependency on A or B being done with virtual packages, as your bq27x00-chip-info example. Feb 04 03:50:12 But I don't know for sure. Feb 04 03:50:15 Googling. Feb 04 03:51:26 googled already, haven't found straightforward solution... Thanks for checking it, too Feb 04 04:05:27 Estel_: What about syntax like this: 'Depends: foo | bar' Feb 04 04:05:54 I see some examples like that @ http://www.debian.org/doc/debian-policy/ch-relationships.html Feb 04 04:06:10 but I'm not sure if I'm misreading it. Feb 04 04:47:14 thanks, I'll check it Feb 04 04:47:44 MentalistTraceur, I just vaguely remember, that this syntax wasn't working for our apt, but I may remember it wrong, or it might have been fud Feb 04 04:47:51 ~bnf Feb 04 04:47:52 hmm... bnf is Backus Naur Form, a language for creating definitions for computer languages and other command sets. See the Jargon File for a good description of how it works. Feb 04 04:48:02 lol Feb 04 04:48:07 ~BNF Feb 04 04:48:08 i guess bnf is Backus Naur Form, a language for creating definitions for computer languages and other command sets. See the Jargon File for a good description of how it works. Feb 04 04:49:33 infobot, no, #maemo bnf is http://talk.maemo.org/showthread.php?p=1211656 Feb 04 04:49:33 Estel_: okay Feb 04 04:49:38 ~bnf Feb 04 04:49:38 extra, extra, read all about it, bnf is http://talk.maemo.org/showthread.php?p=1211656 Feb 04 04:50:06 infobot, bnf is also Backus Naur Form, a language for creating definitions for computer languages and other command sets. See the Jargon File for a good description of how it works. Feb 04 04:50:06 Estel_: i already had it that way Feb 04 04:50:20 ~thanks Feb 04 04:50:21 Estel_: bitte Feb 04 04:50:39 anyway, new version of ~bnf got released Feb 04 04:50:45 happy bnf'ing! Feb 04 04:51:08 more proper release (package) if/when autobuilder and repos start working properly Feb 04 04:51:27 ~truecrypt Feb 04 04:51:27 extra, extra, read all about it, truecrypt is http://talk.maemo.org/showthread.php?t=81435 Feb 04 04:52:00 ~mugen-cover-mod Feb 04 04:53:20 ~mugen-cover-mod is http://talk.maemo.org/showpost.php?p=1038196&postcount=799 Feb 04 04:53:21 okay, Estel_ Feb 04 05:01:43 good night, it's 6AM here Feb 04 06:07:18 if i find the deb in devel, can i just scp it to my n900 and dpkg -i it? Feb 04 06:07:38 or is that not recommended Feb 04 06:10:26 kerio, there is something wrong with your reiplementation of battery applet Feb 04 06:10:44 while it shows capacity from bq, it still seems to compare against design/capacity Feb 04 06:11:34 so, when my 2635 mAh battery is charged to 1251, it shows 47% charged, yet, full green battery icon Feb 04 06:11:51 and on every plug disconnect, even shows "battery charged" banner Feb 04 06:12:07 of course, when plugged, it charges OK, despite banner message Feb 04 06:12:16 (I'm on Pali's bme replacement bits already) Feb 04 06:38:16 http://paste.debian.net/231555/ any ideas? Feb 04 07:49:06 damo22: is the .deb corrupted? Feb 04 07:59:36 thedead1440: i dont know, is there a md5sum built in? Feb 04 08:00:09 one benefit of RPM over DEB i guess Feb 04 08:00:18 :P Feb 04 08:01:17 i enabled extras-devel as well as extras is that bad? Feb 04 08:01:35 i think the deb i downloaded manually is dogpoo Feb 04 08:02:23 the md5sum: 122e064408e3b0060dce987a9b5e7b40 Feb 04 08:08:34 Estel_: i suppose that it would be better to use the bq27k percentual too Feb 04 08:08:57 or possibly the bars Feb 04 08:10:41 damo22: extras-devel is a superset of extras Feb 04 08:10:54 it's not harmful to keep both enabled, it's just useless Feb 04 08:17:44 kerio: that is good to know Feb 04 08:28:32 Estel_: would you be content with something that only uses bq and nothing else? Feb 04 08:36:22 i'm still kind of shocked that the n900 eats battery significantly faster than my 5230 nuron... (the n900 commandeered the nuron's bl5j...) Feb 04 08:38:26 Bear in mind that the 5230 is a much simpler phone Feb 04 08:38:30 the bl5j is old so I don't expect long life but going a few days without using/charging isn't happening Feb 04 08:38:34 which doesn't run anywhere near as much stuff Feb 04 08:39:53 it is indeed... just shocked how much more this uses... Feb 04 08:41:55 ecc2g: with the screen on, it uses about 180mA Feb 04 08:42:25 even with a kickass battery, you've only got 7 hours with the screen on, pretty much Feb 04 08:43:42 I probably should get another cell. I stole the stock nokia branded and a japod from my nuron. Feb 04 08:46:29 what's the 180mA breakdown like? I hope it isn't mostly the backlight... Feb 04 08:47:40 the transflective display is nice, probably contributes to backlight power... Feb 04 08:48:23 stupid nuron completely washes out in sunlight Feb 04 08:53:14 ecc2g: 7mA for cell, 170 for backlight? Feb 04 09:09:54 Estel_: got a deb for you Feb 04 09:10:38 Estel_: https://dl.dropbox.com/u/1405219/status-area-applet-battery_1.0-3%2Bkerio1_armel.deb Feb 04 09:10:44 source at https://dl.dropbox.com/u/1405219/status-area-applet-battery_1.0-3%2Bkerio1.tar.gz Feb 04 09:11:01 seems to work, bq27x00 is the only source of data now Feb 04 09:11:07 and it'll fail horrendously if it's not calibrated Feb 04 09:24:57 hey kerio Feb 04 09:25:02 sup Feb 04 09:25:08 did you calibrate that orange battery? Feb 04 09:25:08 you betrayed me Feb 04 09:25:11 yep Feb 04 09:25:15 and? Feb 04 09:25:18 1400 Feb 04 09:25:27 they why the F does my one say 1600? Feb 04 09:25:34 how the F did you calibrate it? Feb 04 09:25:45 bq27x00 script Feb 04 09:25:55 you're supposed to use constant-ish discharge when calibrating Feb 04 09:26:39 well fuck. Feb 04 09:27:09 I calibrated by leaving internet streaming radio playing with the screen on max brightness. Feb 04 09:27:24 that is about as constant as you can get. Feb 04 09:27:54 newp Feb 04 09:28:15 also oh, it was a higher load than me Feb 04 09:28:21 WHY DO YOU HAVE A BETTER BATTERY YOU ASSHOLE Feb 04 09:31:19 anyway, use a calibration script, calibrate overnight and a constant load Feb 04 09:31:23 like, just the backlight Feb 04 09:55:56 How does warranty work with mobile phones? Feb 04 09:56:07 The N900 is full of bugs. Feb 04 09:56:14 Nokia said they won't fix it. Feb 04 09:56:20 Can they just do that? Feb 04 09:56:50 IMHO, they created a research product, which shouldn't have been distributed to consumers. Feb 04 09:57:02 mm. they can :S Feb 04 09:57:03 Perhaps all the manufacturers do the same thing. Feb 04 09:57:28 What are you buying when you buy a piece of electronics then? Feb 04 09:57:41 Its a war to make them fix the expired cert....what about fixing other things Feb 04 09:58:06 Because, if they can, what stops them from just bricking the device one day after warranty? Feb 04 09:58:16 They could just say it was a 'mistake'. Feb 04 09:59:04 The best thing nokia can do (witch i doubt they'll do ) is to relase the control over maemo to the community . Feb 04 09:59:23 btw are u using CSSU ? Feb 04 09:59:31 I am. Feb 04 09:59:52 What issues do you have with the phone? Feb 04 09:59:57 But still, I consider that I have to use CSSU a failure on Nokia's part. Feb 04 10:00:09 ShadowX: I wanted to have Cell Broadcast working. Feb 04 10:00:11 Except the hildon lagging quite a lot after few days of work, i'm pretty happy with it. Feb 04 10:01:16 I think N900 is the least of their problems atm ..they don't even care about it :) Feb 04 10:01:23 ShadowX: if an asteroid is going to destroy my area in 30 minutes, and I am working, I will be the only one who is going to be incinerated. Feb 04 10:01:54 ShadowX: a Nokia 3210 did support this feature. Feb 04 10:02:11 ShadowX: the N900 also has it, (and someone patched the binary already). Feb 04 10:02:21 Oo Feb 04 10:02:23 fasta: what sort of bugs? Feb 04 10:02:35 i dont see how can you put n900 next to 3210 .... Feb 04 10:02:52 jacekowski: https://bugs.maemo.org/show_bug.cgi?id=8347 Feb 04 10:02:53 04Bug 8347: Cell Broadcast Feature not available Feb 04 10:03:00 that's not a bug Feb 04 10:03:06 jacekowski: yes, it is. Feb 04 10:03:06 that's unsupported feature Feb 04 10:03:17 jacekowski: the code is there and it doesn't work. Feb 04 10:03:19 android phones have no CB support as well Feb 04 10:03:58 Perhaps this is a part of evolution where smartphone owners are supposed to die before the others. Feb 04 10:04:35 "We exterminate all our users. It's part of our new business strategy. " -- Nokia & Android sellers Feb 04 10:05:03 CB is one of those useless features that nobody really supports Feb 04 10:05:14 not many carriers send CB messages Feb 04 10:05:21 jacekowski: lots of countries use them for emergency features. Feb 04 10:08:38 It's like 3GPP videocalls, instant messaging, video/image sharing and VoIP.. There's no widespread usage of any of them, but there's always like one random operator/country where they're used :/ Feb 04 10:08:51 It would make great marketing material for someone who does support it. Feb 04 10:09:24 voip or "walkie talkie mode" Feb 04 10:09:38 that was supported here :P Feb 04 10:15:33 fasta: not really Feb 04 10:15:43 fasta: you may just as well send normal text message to everyone Feb 04 10:15:58 or the one that just pops up on the screen instantly Feb 04 10:16:07 CB has to be configured on the device Feb 04 10:16:20 jacekowski: no, because that require a user to be registered to a cell tower. Feb 04 10:16:36 CB requires user to be subscribed to specific channels Feb 04 10:16:46 jacekowski: and you have to know who is in the given area. Feb 04 10:17:04 CB does the same Feb 04 10:17:09 jacekowski: knowing who is in a given area has privacy problems and data storage problems. Feb 04 10:17:29 as in, phone will not recieve message if it's not subscribed Feb 04 10:17:34 because it will not listen for them Feb 04 10:17:55 so it's from different side, but effect is the same Feb 04 10:18:03 fasta: they do know where everyone is Feb 04 10:18:27 fasta: they have to store that and be able to look it up quickly so they know where to send calls for you Feb 04 10:18:53 jacekowski: yes, but to send a CB you don't need position. Feb 04 10:19:35 yes, but tower knows all registered users anyways Feb 04 10:20:29 so you send message to the tower and that then sends it to all registered handsets Feb 04 10:20:33 same effect Feb 04 10:21:00 sorry for delayed FW reboot. I improved my WD script now, hope this will trigger more early from now on Feb 04 10:21:14 where i can find openntp settings? Feb 04 10:21:17 jacekowski: how does the tower know that some client of a competing company is in his area? Feb 04 10:21:28 fasta: it doesn't Feb 04 10:21:35 fasta: but phone will not recieve that message anyways Feb 04 10:21:50 fasta: phone only listens to CB from tower it's registered to Feb 04 10:21:51 jacekowski: are you saying that CB is provider specific? Feb 04 10:21:54 yes Feb 04 10:22:01 jacekowski: that's a rather broken idea. Feb 04 10:22:09 jacekowski: who comes up with such bad ideas? Feb 04 10:22:12 exactly Feb 04 10:22:28 that's why i'm saying, CB has no benefit over sending same text to multiple users Feb 04 10:22:50 lower bandwidth Feb 04 10:23:01 jacekowski: but if the SMS network is crowded, isn't there some problem? Feb 04 10:23:18 yes and no, if it's emergency message it will get prioritised anyways Feb 04 10:23:24 so that's not a problem Feb 04 10:23:40 jacekowski: ok, I think I get it. Feb 04 10:24:20 our top download IP on rmo leeching with a awesome 459kBps Feb 04 10:24:26 First CB was created for some unreleated applicatio, then some government thought it was a good idea to use it for ER (which it isn't), and now I have a problem. Feb 04 10:25:07 What should have happened, was the government creating a special standard for ER which was independent of the provider. Feb 04 10:25:22 extremly power inefficient Feb 04 10:25:34 and impossible to implement in a phone Feb 04 10:25:47 unless you want to go back to brick sized phones Feb 04 10:26:02 jacekowski: the problem being that listening forever takes a lot of energy? Feb 04 10:26:02 phone can only listen on one channel at any given time Feb 04 10:26:19 ie. channel of the tower it's registered to Feb 04 10:26:24 jacekowski: can't it just connect to a circuit which is powered by the signal? Feb 04 10:26:41 no Feb 04 10:26:42 jacekowski: so, when a very high energy signal comes in, it's provides power for the logic to process it? Feb 04 10:26:44 I'd be tempted to kill such leecher connections Feb 04 10:27:05 DocScrutinizer05: who is it? Feb 04 10:27:06 fasta: brick size Feb 04 10:27:06 My carrier sends CBSMS to my phone Feb 04 10:27:23 probably we should do SPI and reject everything that has no valid maemo apt browser ID Feb 04 10:27:26 jacekowski: is there some proof of that? Feb 04 10:27:26 although only for cell tower names Feb 04 10:27:29 DocScrutinizer05: holy shit rmo is blazing fast Feb 04 10:27:34 did warfare do something? Feb 04 10:27:39 nah Feb 04 10:27:42 fasta: that sort of hardware has been in use 50 years ago Feb 04 10:27:46 not that I'd know of Feb 04 10:27:56 well, we probably did some throttling Feb 04 10:27:56 DocScrutinizer05: srsly, it's... responsive, at least Feb 04 10:27:58 i'm not sure about fast Feb 04 10:28:04 fasta: radio recievers powered by radio waves Feb 04 10:28:14 fasta: but those required 100m long aerials to operate Feb 04 10:28:28 jacekowski: where do you learn this stuff? Feb 04 10:28:30 fasta: and transmitters that were 100x as powerfull as GSM Feb 04 10:28:33 kerio: less clients updating things. Feb 04 10:28:36 yay Feb 04 10:28:37 fasta: wikipiedia Feb 04 10:28:44 warfare: so... thanks for nothing! :D Feb 04 10:28:52 jacekowski: not part of standard EE university curriculum? Feb 04 10:28:53 root@stage:~# netstat -tn|wc -l Feb 04 10:28:54 2097 Feb 04 10:28:56 what's the current throughput? Feb 04 10:28:57 \o/ Feb 04 10:29:05 kerio: still 25 mbps. Feb 04 10:29:05 same as it ever was Feb 04 10:29:16 (and we should still implement some sort of CDN) Feb 04 10:29:24 fasta: nope Feb 04 10:29:41 fasta: http://en.wikipedia.org/wiki/Crystal_radio Feb 04 10:30:07 fasta: but the thing is, it had circuitry tuned to specific frequency Feb 04 10:30:35 jacekowski: well, can it be physically adjusted in a simple way? Feb 04 10:30:36 I do find it cool that almost every crazy idea I come up with has been thought of already. Feb 04 10:30:42 yes Feb 04 10:30:52 but it's still one frequency Feb 04 10:30:52 like, a slider or a dial Feb 04 10:30:57 oh, of course Feb 04 10:31:02 do you know the most simple spy bug? two wires with a charcoal grain telephone mic in the middle Feb 04 10:31:05 fasta: and someone already said that :D Feb 04 10:31:06 so you can't do it between towers Feb 04 10:31:17 unless in emergency every tower would transmit on same frequency Feb 04 10:31:21 but then you have intereference problem Feb 04 10:31:28 kerio: you mean what I just said? :) Feb 04 10:31:33 yep Feb 04 10:32:03 Pali: the battery applet code is somewhat of a mess :S Feb 04 10:32:11 I think it is extremely hard to actually know what the rest of the world has already done at this point. Feb 04 10:32:24 so doing emergency broadcasts is hard Feb 04 10:32:27 I.e., knowing of all the patents in existence. Feb 04 10:32:39 (rmo) well I asked teotwaki to stop mirroring, might actually have helped Feb 04 10:32:48 (and some things which are likely trade secrets) Feb 04 10:32:55 Pali: also, what's with bme_running and the fact that it's only set for one cycle? Feb 04 10:33:02 aiui Feb 04 10:33:39 DocScrutinizer05: don't you also need a battery for that? Feb 04 10:33:43 now blacklist that leecher on -224-25-100.tica.com.ua. and get another 500kBps bandwidth ;-D Feb 04 10:34:03 DocScrutinizer05: so you only have 25MBps? Feb 04 10:34:07 fasta: no, two wires and a passive microphone built from charcoal Feb 04 10:34:16 DocScrutinizer05: and what kind of equipment do you need to receive that? Feb 04 10:34:22 jacekowski: yep, seems so Feb 04 10:34:29 fasta: something with power, i assume :) Feb 04 10:34:35 DocScrutinizer05: i can do some reverse proxying and stuff if you are interested Feb 04 10:34:42 i've got gigabit of bandwidth Feb 04 10:34:51 fasta: a strong transmitter with directional antenna, and a RX with directional antenna Feb 04 10:34:57 jacekowski: warfare said he has a server too Feb 04 10:35:04 the problem is that the n900's apt doesn't do redirects Feb 04 10:35:12 jacekowski: how much traffic do you have? ;) Feb 04 10:35:18 100T Feb 04 10:35:19 so we'd have to change the dns records, but the dns records are still controlled by nokia Feb 04 10:35:21 DocScrutinizer05: are you a consultant to spy agencies? :) Feb 04 10:35:35 jacekowski: the problem is we don't have any control over DNS yet Feb 04 10:35:36 DocScrutinizer05: what do you have to transmit? Feb 04 10:35:44 fasta: yes Feb 04 10:35:47 ;-P Feb 04 10:35:56 DocScrutinizer05: that's not helping for sure Feb 04 10:36:05 kerio: a carrier, nothing else Feb 04 10:36:07 DocScrutinizer05: yeah, the problem is that you probably wouldn't be allowed to say it if you were. Feb 04 10:36:26 DocScrutinizer05: has board given an ETA for the domain ownership transfer? Feb 04 10:36:26 the two wires will tune on it by length, and the mic AM modulates it Feb 04 10:36:33 no Feb 04 10:36:40 has board given an ETA for anything? Feb 04 10:36:43 jacekowski, maybe teotwaki could mirror from you though.. and maybe rmo could have rsync service so that you could get new stuff efficienctly Feb 04 10:36:46 no Feb 04 10:36:58 DocScrutinizer05: is there some book which describes this in tutorial fashion? Feb 04 10:37:01 fasta: since he said "yes", then he's not a consultant for a spy agency Feb 04 10:37:08 unless that's what he wants you to believe Feb 04 10:37:12 err yes, Nokia sponsoring ending Feb Feb 04 10:37:16 end of Feb 04 10:37:28 is an ETA board has conbfirmed lately Feb 04 10:37:30 DocScrutinizer05: clearly, they can't keep sponsoring our dns :P Feb 04 10:38:37 Is there any way you could trick some low level chump into giving you access to the fremantle closed bits? Feb 04 10:38:53 ShadowJK: thanks that someone else brings rsync in again Feb 04 10:38:55 vi____1: the password is swordfish Feb 04 10:39:17 did u hack that on a gameboy? Feb 04 10:39:43 i wrote a GUI in visual basic for it Feb 04 10:39:46 PS3 from Iran. Feb 04 10:40:10 It comes with a monkey these days. Feb 04 10:40:15 chem|st, i dont think we can convince people to not try to sync and mirror, so we might as well give them efficient access.. Feb 04 10:40:27 There is a wrong monkey on the box sometimes, though. Feb 04 10:40:45 I maekd a .bat virus to get the c0des from a injection worm lizard ppl 9/11 was a hologram. Feb 04 10:40:52 ShadowX: ++ Feb 04 10:40:55 ShadowJK: what I said... Feb 04 10:41:06 Just put up a daily torrent. Feb 04 10:41:17 fasta, that's inefficient Feb 04 10:41:26 Because they'd be redownloading everything daily Feb 04 10:41:30 instead of just downloading changes Feb 04 10:41:35 ShadowX: with diffs Feb 04 10:41:44 heh Feb 04 10:42:13 It's in the interest of the leecher to download as little as possible. Feb 04 10:42:24 ShadowJK: haha, sure we would *love* to give _everybody_ lightning fast access Feb 04 10:42:39 DocScrutinizer05: are you the only one who can visit it now? ;) Feb 04 10:43:01 DocScrutinizer05, it's about cutting down on bytes transfered by these mirroring people Feb 04 10:43:19 ShadowJK: apt-mirror is somewhat efficient Feb 04 10:43:24 it only downloads the changed packages Feb 04 10:43:33 fasta: the only one except warfare you will ever talk to, I guess Feb 04 10:43:48 And then if we could convince them to donwload from jacekowski offered bandwidth it'd be even better :-) Feb 04 10:44:09 ShadowJK: it's just the first update, anyway Feb 04 10:44:15 extras doesn't change a lot Feb 04 10:44:23 and it's the main culprit, really Feb 04 10:44:36 together with extras-devel, which already has some mirrors that are better than the original Feb 04 10:44:59 SPI with reject for everyone not using "APT maemo" browser Feb 04 10:45:12 DocScrutinizer05: why? :s Feb 04 10:45:28 what's the index page for, then? Feb 04 10:45:35 will relatively effectively reject mirrors and other leechers Feb 04 10:45:53 DocScrutinizer05: until they change their user agent :) Feb 04 10:46:11 proper traffic shaping for those non-APT-maemo connections would be even better Feb 04 10:46:22 kerio: sure they could do Feb 04 10:46:46 kerio: it however wouldn't change much for them, since then again everything collapses Feb 04 10:46:57 so either way they don't get a proper image Feb 04 10:47:35 kill connections after they downloaded >100MB Feb 04 10:47:48 fail2ban alike Feb 04 10:48:24 "sorry dude! no more than 100MB download per day. Your IP got blocked for 24h" Feb 04 10:50:07 it's funny, I thought everything else but rmo would be the hard stuff Feb 04 10:50:13 and rmo would be the easiest, since it's "only" bandwidth Feb 04 10:50:30 dang, I wonder who's that .ua leecher Feb 04 10:50:32 ShadowJK: as it turns out, we've got more n900s than we thought Feb 04 10:50:38 DocScrutinizer05: what's .ua? Feb 04 10:50:47 nfc Feb 04 10:50:49 ukraine Feb 04 10:50:51 We have 2000Gb/s fiber optical connections. Such a waste of talent that people need to spend time on reducing bandwidth. Feb 04 10:51:07 * ShadowJK has 100M (ish) at home Feb 04 10:51:29 By "we", I meant some research lab. Feb 04 10:51:29 fasta: it's just a workaround for a bureaucratic issue, we could be using cloudflare if we had control of the dns Feb 04 10:51:45 1Gbit proper ISP backbone would already make us happy Feb 04 10:51:47 (there's a free tier for cloudflare, wtf - how do they make money?) Feb 04 10:52:12 kerio: the brand becomes more valuable if it has more users. Feb 04 10:52:12 * ShadowJK wonders if 4chan is on the free tier too Feb 04 10:52:18 kerio: larger users will pay. Feb 04 10:52:21 by shanghaiing your DNS Feb 04 10:52:22 kerio: With the paid contracts. Which you actually want if you run a business on top of a cdn. Feb 04 10:52:35 DocScrutinizer05: is that a technical term? Feb 04 10:52:36 ShadowJK: the first paid tier is 20$/month Feb 04 10:52:52 DocScrutinizer05: what's good about shangaiing dnses? Feb 04 10:52:57 fasta: somewhat Feb 04 10:53:45 kerio: all that we can't do now since we didn't, for 'our' DNS Feb 04 10:54:45 kerio: the ability to decide on your own discretion and being in control and possession of your DNS Feb 04 10:55:07 DocScrutinizer05: well, apparently they're also testing a CNAME solution for subdomains Feb 04 10:55:18 DocScrutinizer05: I understand what the word means; I don't see what you meant, however. Feb 04 10:55:33 and besides, it's the only way you could do CDN for your top level domain without redirects, aiui Feb 04 10:55:42 probably if we ever gain control over maemo.org, the bylaws in contratc with Nokia would even forbid to hand control over it to somebody else Feb 04 10:56:05 DocScrutinizer05: you gain control of the domain when you're the legal entity responsible for it according to the registrar Feb 04 10:56:15 you can set your root nameserver to whatever you want, that's not giving control of the domain Feb 04 10:56:16 exactly Feb 04 10:56:29 kerio: that's what I thought. Feb 04 10:57:52 you've got the magical implied A records for your root nameservers set with the registrar (if you decide to have nameservers in your domain) Feb 04 10:58:08 so you can "bootstrap" your DNSes Feb 04 10:59:19 The only problem I have with DNS is that the software that needs to run it is written by amateurs. Feb 04 10:59:30 kerio: see a few days back in #maemo-ssu (iirc). It's been found that that one free CDN has no PHP but only static content, and they want admin-c over your domain aiui Feb 04 10:59:57 since NS would be theirs Feb 04 11:00:24 yeah but you can take away the NS at any time Feb 04 11:00:40 not if they are admin-c Feb 04 11:00:44 afaik Feb 04 11:00:56 oh, i see Feb 04 11:01:00 yeah, that's bullshit :) Feb 04 11:01:14 coralcdn is also free, but it only works with redirects Feb 04 11:01:26 which won't work, because the stock apt is older than dirt Feb 04 11:02:35 can we create a maemocdn.org domain, point its NS over to cloudflare, then in maemo.org CNAME repository to repository.maemocdn.org? :-) Feb 04 11:02:49 maybe we can build our own cdn - setup powerdns with geomatching and have community supported varnish boxes ;) Feb 04 11:03:01 only feasible way: we stay owner and admin-c of maemo.org DNS, but we redirect to CDN's NS in our DNS Feb 04 11:03:09 ShadowJK: neat Feb 04 11:03:15 s/redirect/point/ Feb 04 11:03:16 DocScrutinizer05 meant: only feasible way: we stay owner and admin-c of maemo.org DNS, but we point to CDN's NS in our DNS Feb 04 11:03:21 ShadowJK: i doubt it would work Feb 04 11:03:36 you need the correct hostname request for http to work properly Feb 04 11:04:12 I don't know if APT or HAM handles CNAME Feb 04 11:04:14 Anyway, we do need a DNS host at some point or another, might as well be cloudflare? :-) Feb 04 11:04:18 DocScrutinizer05, everythign does Feb 04 11:04:35 DocScrutinizer05: getaddrbyname handles CNAME :) Feb 04 11:04:51 just tell everyone to use a different repository Feb 04 11:04:53 ? Feb 04 11:04:59 problem solved Feb 04 11:05:04 well, then that's what I already suggested when we can't get control over maemo.org. We need Nokia to create CNAME to domains we own Feb 04 11:05:22 damo22: the whole point of all of this is that we'll eventually gain proper control of maemo.org and then slip in cssu Feb 04 11:05:28 ah Feb 04 11:05:34 possibly with something that replaces the boot video with a static "NOKIA SUCKS" image Feb 04 11:05:45 lol Feb 04 11:05:51 Oh, I vaguely rememberd there was restrictions with CNAMEs, but I forgot what those were. Now I remember, you can only have one CNAME record, but you can have multiple A record. Feb 04 11:06:00 DocScrutinizer05: the people over in #dns generally laugh at you when you use CNAMEs. Feb 04 11:06:18 fasta: are you sure they're not laughing at DNAMEs? Feb 04 11:06:26 I can generally live with that Feb 04 11:06:55 kerio: I haven't heard about DNAMEs. So, I am going to assume they don't exist. Feb 04 11:07:06 if the person who controls the DNS records for maemo.org is currently employed by nokia how are you gonna get control of it... ever Feb 04 11:07:06 [disclaimer] I have a cloudy idea of how DNS stuff works, at best Feb 04 11:07:17 DocScrutinizer05: DNS is non-trivial, IMHO. Feb 04 11:07:22 DocScrutinizer05: when you use CNAMEs for http servers, the final server has to accept that he's supposed to be responding to your CNAME too Feb 04 11:07:24 It's trivial in theory. Feb 04 11:07:29 fasta: it's a good assumption, really Feb 04 11:07:34 they're deprecated and a generally bad idea Feb 04 11:07:42 fasta: yes, that's the idea of CNAME Feb 04 11:07:48 Making it work is something entirely different. Feb 04 11:07:52 oops, Feb 04 11:07:56 kerio: ^^^ Feb 04 11:08:06 DocScrutinizer05: i doubt cloudflare will "accept" that :) Feb 04 11:08:19 kerio: yep Feb 04 11:08:41 do one of you guys here control the DNS zone file? Feb 04 11:08:43 If I ever need personel for a networking position, I know exactly how I am going to grill them :) Feb 04 11:08:51 wait for warfare and what he comes up with eventually Feb 04 11:09:19 he is checking alternatives he knows of by his professional experience, this very minute Feb 04 11:09:31 damo22: i think that the domain is still owned by nokia oyj Feb 04 11:09:34 damo22; nope. We/they have to beg nokia for any changes Feb 04 11:09:53 i see, Feb 04 11:10:04 and until they control the domain, we can't change the name servers for the maemo.org domain Feb 04 11:10:08 Should get as much users to harass Nokia with requests to get their shit straight Feb 04 11:10:13 and the name servers are NS*.NOKIA.COM Feb 04 11:10:20 which are obviously controlled by nokia :) Feb 04 11:10:50 one of you guys need to apply for a job with nokia and gain control of the admin position Feb 04 11:10:56 :P Feb 04 11:11:20 then by mistake, set the nameservers to Feb 04 11:11:24 when you're hired by nokia, you get a chip implanted in your brain Feb 04 11:11:29 and nowadays, that chip runs windows CE Feb 04 11:11:31 http://wiki.maemo.org/Migrating_to_Community-driven_Infrastructure#Who_owns_.22maemo.org.22.3F Feb 04 11:11:39 Windows8, mind you:) Feb 04 11:11:44 RTFM ;-P Feb 04 11:11:51 the CE based version is deprecated and looking at the 7.8 update its a joke too Feb 04 11:12:24 if M$ ever tries to install DRM enforcing brain chip kill me okay? Feb 04 11:12:37 MS will try that just as hard as any other company Feb 04 11:13:04 Amusingly I trust MS more than Google now :-) Feb 04 11:13:07 SecureBoot will prevent any linux os from booting on it Feb 04 11:13:19 MS has some really friendly FOSS terms for their appstore Feb 04 11:13:24 ShadowJK: really?? why Feb 04 11:13:38 damo22; wrt privacy and such, I mean Feb 04 11:13:46 I trust Google as much as I trust MS Feb 04 11:13:50 which is not at all Feb 04 11:13:59 i trust a big company as far as i can throw them Feb 04 11:14:05 exactly Feb 04 11:14:11 i'd like a brain chip to access wireless networks Feb 04 11:14:20 except that i also read reflections on trusting trust Feb 04 11:14:35 and i'm not good enough at reading circuits to figure out if the actual hardware is trustworthy Feb 04 11:14:49 yeah that is going to be a problem Feb 04 11:15:30 It would be great if we all had chips in our head produced by the same company Feb 04 11:15:35 sounds like a really great idea Feb 04 11:15:38 i think the only solution to the digital trust problem is open hardware specs, open software specs everything Feb 04 11:16:04 I still opt for the solution that JC Denton presented at the end of Deus Ex Invisible War Feb 04 11:16:21 so if someone finds a problem with the current implementation, they are free to change it and reimplement it Feb 04 11:16:28 thats all Feb 04 11:16:34 http://www.youtube.com/watch?v=xBeoreJr4Yc Feb 04 11:16:42 damo22: nope, you still have a trust problem for the manufacturing Feb 04 11:17:06 quality control Feb 04 11:17:11 and testing Feb 04 11:17:16 should fix that Feb 04 11:17:17 I'm not sure what company my CC number is safest with. Probably MS. Google probably doesn't give it out, but like paypal it might probably suddenly try charge something random for no good reason. As for Nokia, well, Nokia/Ovi store is so incompetent, that even if I gave them my CC number, they'd probably delete it accidentally, and if not, they're usually not able to figure out how to charge a CC anyway, even when I want them to :-) Feb 04 11:17:42 I would totally trust Nokia with all my personal information indeed Feb 04 11:17:50 They are such a bunch of fuckups that they can hardly pose any danger Feb 04 11:17:58 :S Feb 04 11:18:34 damo22: nope Feb 04 11:18:45 there's no amount of testing that can figure out if someone is planting a hardware bug in your cpu Feb 04 11:19:05 wtf Feb 04 11:19:40 what if the microcode is open source Feb 04 11:19:55 damo22: **manufacturer** Feb 04 11:20:06 or shipping Feb 04 11:20:10 if they wire it up wrong inside, it wont work Feb 04 11:20:13 surely Feb 04 11:20:21 the problem is not that they might wire it wrong Feb 04 11:20:28 the problem is that they might wire it /just right/ Feb 04 11:20:46 according to the specs, a cpu instruction should perform an action that is expected, every instruction can be tested Feb 04 11:20:58 and your testing will reveal absolutely nothing wrong Feb 04 11:21:12 bullshit Feb 04 11:21:28 Who is going to say the Chinese havent done so already Feb 04 11:21:36 Lava_Croft: that's the whole point! Feb 04 11:21:41 yup Feb 04 11:21:51 Anyone who played the original Deus Ex knew this already:) Feb 04 11:21:53 hang on a sec Feb 04 11:21:58 damo22: nope, you basically can't test anything inside a SoC or CPU Feb 04 11:22:16 Lava_Croft: you linked us a video of deus ex invisible war, you have to hand in your deus ex fanboy card Feb 04 11:22:42 you can check if the results are like you expected, but you won't find out about a phantom CPU that's doing code inspection and eventually intercepts/takes over Feb 04 11:23:03 thats a bit paranoid Feb 04 11:23:08 wouldnt you say Feb 04 11:23:11 sure, but feasible Feb 04 11:23:34 btw RISC has no microcode afaik Feb 04 11:23:53 just because you're paranoid, it doesn't mean that they're not out to get you ;) Feb 04 11:24:05 i know Feb 04 11:24:15 some modern intel cpus can be "upgraded" via software Feb 04 11:24:23 :O Feb 04 11:24:33 wth Feb 04 11:24:40 upgrade a cpu??? Feb 04 11:24:53 yeah, if you buy the low-end model you don't get the cool virtualization features Feb 04 11:24:54 sure, why not Feb 04 11:24:59 damo22: microcode updates Feb 04 11:25:02 :nod: Feb 04 11:25:06 this is bad Feb 04 11:25:07 you can buy a card with a code, and then you unlock those features with your code Feb 04 11:25:18 defective products by design Feb 04 11:25:24 defective for who? Feb 04 11:25:27 way cheaper that to produce different chips Feb 04 11:25:28 for the user Feb 04 11:25:32 definetely not for the people who're out to get you Feb 04 11:25:35 s/that/than/ Feb 04 11:25:35 DocScrutinizer05 meant: way cheaper than to produce different chips Feb 04 11:25:35 >:D Feb 04 11:26:02 give us a super chip that can be i7 but locked down to i3 Feb 04 11:26:24 that's a very common practive with appliences already Feb 04 11:26:25 bastards Feb 04 11:26:38 neutered cpus :< Feb 04 11:26:38 from wafer edges you get i3 from wafer center you get i7, i5 is afair another design Feb 04 11:27:16 the cheap TV also has all the cute capabilities to do HDD recording etc, just locked in firmware Feb 04 11:27:19 all this quad tripple dual core things is one chip Feb 04 11:27:45 :nod: Feb 04 11:28:03 DocScrutinizer05: someties they just not install required hardware Feb 04 11:28:09 defective cores get either turned off by software and may be unlocked or by Focused Ion Beam cut out -> not to be unlocked Feb 04 11:28:37 microcode upgrades are different thing4 Feb 04 11:28:44 microcode updates are there to fix bugs Feb 04 11:28:56 they produce flash; they design it for 34GB and sell as 32. Now if a chip is too damaged to get the 32 but has only 30, they lock it down to 16 Feb 04 11:29:00 jacekowski: I thought it is the same codebase for unlocking Feb 04 11:30:04 same with quad core CPU Feb 04 11:30:21 DocScrutinizer05: wtf, flash is so cheap they can just bin the chip.. Feb 04 11:30:53 one core not complying with the speed stepping, they rather lock that one core than rating down speed stepping for all 4 cores Feb 04 11:31:23 jacekowski: enlighten please Feb 04 11:38:12 one day it will be the same with software Feb 04 11:38:34 all the features built into the super model, modules switched off for lower end Feb 04 11:38:47 chem|st: upgrade thing is completly different set of features Feb 04 11:38:58 chem|st: microcode is there to patch existing opcodes in cpu Feb 04 11:39:17 chem|st: and microcode is horribly slow Feb 04 11:39:24 its probably not a bad thing, but its bad if they do it in a way that is anticompetitive Feb 04 11:39:44 like creating a monopoly on the app store for example Feb 04 11:40:20 jacekowski: ah ok now I get what you mean Feb 04 11:40:21 chem|st: but slow instruction is better than incorrectly working instruction Feb 04 11:40:54 they are probably working on a cpu that has instructions 'locked', and only verifies itself with software Feb 04 11:41:04 and serial number of registered software Feb 04 11:41:13 :( Feb 04 11:41:48 software will be hacked... Feb 04 11:42:05 yes it will, but it will get more and more difficult Feb 04 11:42:55 since to get a development environment will be tricky Feb 04 11:42:56 do you know why people do not hack the pentagon that often? Feb 04 11:43:08 because there's not much of interest in it? Feb 04 11:43:14 that too Feb 04 11:43:22 because there are people on the other side of the firewall Feb 04 11:43:27 that know what they are doing Feb 04 11:43:34 and "monitor" it constantly Feb 04 11:43:36 jacekowski: exactly Feb 04 11:43:43 jacekowski: meh, people that know what they're doing isn't really the problem Feb 04 11:43:49 the problem are their friends with the $big_gunz Feb 04 11:44:15 why did this convo turn to the pentagon Feb 04 11:44:21 we are talking about cpus Feb 04 11:44:31 damo22: it's because they control the chip that's currently inside your brain Feb 04 11:44:40 what chip? u suck Feb 04 11:44:42 kerio: it takes about the same time to reroute your connection to be hidden as it takes them to follow the route Feb 04 11:44:59 damo22: that's what the chip makes you think Feb 04 11:45:03 LOL Feb 04 11:45:53 im sure theyd love to do that, but they dont need it, they have facebook and iphones Feb 04 11:46:10 hehe Feb 04 11:47:14 a teacher asked students about what STASI wanted to know about their people, then he let the students strike out anything not on facebook Feb 04 11:47:51 the teacher used to be my neighbour Feb 04 11:48:25 there is a level of balance between caring about society and caring about yourself, we all need to make a living Feb 04 11:49:33 too little care about society makes you a good businessman and an arsehole, too much care about society and you become a mental nutcase who cant make a living Feb 04 11:50:24 i like to keep the balance in the middle Feb 04 11:50:55 I like to be the bad guy... Feb 04 11:51:12 this is the reason i contribute to free software Feb 04 11:51:40 when i can Feb 04 11:52:50 and i love the n900 because it gives me the freedom to run a free operating system (with a few blob exceptions) Feb 04 11:54:36 if i had the luxury of not having to make a living, i might take the point of view that i should reject any technology that is restricted in any way at all Feb 04 11:57:16 root@stage:~# grep "91.22x.2x.10x" /var/log/varnish/varnishncsa.log|wc -l Feb 04 11:57:17 2565 Feb 04 11:57:19 root@stage:~# wc -l /var/log/varnish/varnishncsa.log Feb 04 11:57:20 407609 /var/log/varnish/varnishncsa.log Feb 04 12:00:03 188.117.59.205 => 91.22x.2x.10x 0.99Mb 638kb 600kb Feb 04 12:00:56 91.22x.2x.10x - - [04/Feb/2013:06:34:18 +0000] "GET http://repository.maemo.org/extras-devel/pool/fremantle/free/source/j/jammo7-12-data/jammo7-12-data_0.7.24-1.dsc HTTP/1.1" 200 696 "-" "Wget/1.14 (linux-gnu)" Feb 04 12:01:15 what is jammo? Feb 04 12:01:22 some app Feb 04 12:01:39 that's just one example out of the 2565 Feb 04 12:02:13 somebody mirroring the repo with wget Feb 04 12:02:31 taking 1/25 of available bandwidth Feb 04 12:02:42 there are others like him Feb 04 12:02:48 geez Feb 04 12:03:02 so? ban him Feb 04 12:03:15 so hes using multiple ips to mirror the whole repo? Feb 04 12:03:21 kerio: I'm discussing it with warfare Feb 04 12:03:28 temporarily, of course Feb 04 12:03:37 we could also decide that right now, it's fast enough Feb 04 12:03:39 damo22: why do you think that? Feb 04 12:04:00 who do we know from ukraine? Feb 04 12:04:08 :shrug: Feb 04 12:04:17 i dont know Feb 04 12:04:27 E*_ is polish, fmg is bulgarian Feb 04 12:04:28 i thought the x;s represented multiple numbers Feb 04 12:04:52 the x represent privacy protection introduced by me Feb 04 12:05:23 i see, PCI-DSS-IP Feb 04 12:05:25 DocScrutinizer05: we could make jacekowski use his multi-gbit connection to flood him :D Feb 04 12:05:27 also called censoring ;-) Feb 04 12:05:48 lol Feb 04 12:06:06 kerio: nice idea to throttle that client ;-P Feb 04 12:06:08 multi-gbit? :D Feb 04 12:06:24 is he a tier 1 node? Feb 04 12:06:34 apparently Feb 04 12:06:50 :-) Feb 04 12:06:51 modified fmtxd binaries at the speed of light! Feb 04 12:07:00 or you could publish his ip and we can nmap him and play with his ports Feb 04 12:07:13 :P Feb 04 12:07:15 damo22: at some point, it becomes illega Feb 04 12:07:16 l Feb 04 12:07:23 that's exactly why I don't publish full IP Feb 04 12:07:37 pinging him... eeh Feb 04 12:07:51 DDoS-ping? Feb 04 12:08:08 odds are this is a benevolent dude running a maemo mirror Feb 04 12:08:42 maybe even on his company's server Feb 04 12:09:15 many users don't follow TMO or IRC Feb 04 12:09:26 we might cause DDoS on a "innocent" infra Feb 04 12:09:40 it might be a script that backs up the mirror Feb 04 12:09:45 and there are probably still active non-english N900 communities Feb 04 12:10:02 so they might be running their own mirrors once they found the repos work again Feb 04 12:10:21 might ruin that dude's life, getting fired for abusing his company's IT Feb 04 12:10:30 can i purchase a DVD set of Maemo extras? Feb 04 12:10:38 ahah Feb 04 12:10:42 this isn't debian Feb 04 12:10:48 sure, I'll send you for 100EUR ;-D Feb 04 12:10:50 it'd be obsolete before they're shipped Feb 04 12:10:53 :P Feb 04 12:11:18 send it to you* Feb 04 12:11:40 how many Gb is the full repo? Feb 04 12:11:52 good question Feb 04 12:12:05 depends on what you include to "full" Feb 04 12:12:11 hmm Feb 04 12:12:25 some 1.x TB iirc the stats of skeiron Feb 04 12:12:37 i see Feb 04 12:12:56 DVD set isnt a good option then Feb 04 12:13:05 USB HDD maybe Feb 04 12:13:11 tapes :) Feb 04 12:13:17 :nod: Feb 04 12:13:25 DocScrutinizer05: I told you I'd be having fun Feb 04 12:13:27 http://i.imgur.com/WKMTWbv.png Feb 04 12:13:48 err Feb 04 12:13:52 eddyb: OMG i have seen that screen before with nouveau on my MBP Feb 04 12:13:55 .oO(???) Feb 04 12:14:13 that's scanning for ARM code (the probably most common subsequence between all ARM instructions, the ALWAYS condition code) Feb 04 12:14:57 sorry, this picture reminds me to "snowing at nighttime" Feb 04 12:16:01 a somewhat-continuous line of the same color indicates ARM code Feb 04 12:16:30 the beginning of the larger white section at the bottom is a function calling other functions with its parameters Feb 04 12:16:36 s/params/args Feb 04 12:19:26 all the effort people put into RE, they could have made a new one from scratch Feb 04 12:20:29 damo22: cmt checks a signature Feb 04 12:20:59 cmt=? Feb 04 12:21:27 the phone inside the n900 Feb 04 12:21:31 eddyb: what exactly are you trying to reverse engineer and why? Feb 04 12:21:57 cellular modem firmware I get but which bit and why do you care about it? Feb 04 12:22:22 and most importantly, do you have a copy of the hex-rays arm disassembler? Feb 04 12:22:32 hex-rays? Feb 04 12:22:35 that joke? Feb 04 12:22:49 yes I've used it once Feb 04 12:22:50 I have never used hex-rays on ARM Feb 04 12:22:58 but hex-rays does a good job on x86 Feb 04 12:23:11 I don't want a good job, I want an excellent job Feb 04 12:23:15 can u write a disassembler with the spec sheet of the cpu instruction set? Feb 04 12:23:26 http://privatepaste.com/4cf2beca95 Feb 04 12:23:37 (some connection speeds) Feb 04 12:24:03 ok, so what exactly are you reverse engineering this for? Feb 04 12:24:34 if you just want to talk to the cellular modem, there are easier ways to do that than reverse engineering its firmware :) Feb 04 12:24:54 writing a replacement isn't going to happen because of the security on the cellular modem Feb 04 12:24:59 damo22: not sure what you're getting at, but I have a decompiler in the works, and for disassembly, I almost copy-pasted the opcode table in the intel manuals Feb 04 12:25:13 jonwil: yeah, I know all the security issues. but I'm curious Feb 04 12:25:30 i imagine if you have hexdumps of firmware you can read off the instructions and map them to opcodes Feb 04 12:25:43 yeah Feb 04 12:25:47 but if you look at that image Feb 04 12:25:54 only small pieces of it is ARM Feb 04 12:26:04 oh? Feb 04 12:26:05 someone told me to look at C6 Feb 04 12:26:12 err, enter key is happy today Feb 04 12:26:21 C64x+ Feb 04 12:26:24 Couldn't some of it be for a DSP chip? Feb 04 12:26:34 hmmm yeah C64 is a DSP Feb 04 12:26:51 jonwil: yeah, that's what he said. the header actually confirms that about half the binary is for the DSP Feb 04 12:27:04 warfare invents some conditional blacklisting based on criteria from /var/log/varnish/varnishncsa.log Feb 04 12:27:30 it might be worth converting the hexdump into ascii and checking for strings Feb 04 12:27:49 DocScrutinizer05: looks good, apart for the two unlucky bastards who got 5kb/s Feb 04 12:27:54 I suggested sth along the lines of only allowing a few hundered wgets per day and distinct IP Feb 04 12:28:10 C64's instructions are similar to ARM, but the ALWAYS condition code is 0, so it's going to be harder to look for them Feb 04 12:28:28 DocScrutinizer05: what's the problem, at the moment? Feb 04 12:28:33 damo22: there's actually parts of the C code embedded in the binary Feb 04 12:28:41 left in there probably by ASSERT Feb 04 12:28:44 kerio: repo slow? Feb 04 12:28:45 :) Feb 04 12:28:48 that seemed bizarre until I realized C64 is not commodore 64 Feb 04 12:28:48 DocScrutinizer05: is it slow? Feb 04 12:29:40 jonwil: this would be so much easier if the lengths in the header wouldn't be amounting to almost double the file size Feb 04 12:30:40 eddyb: so are there function calls that jump to raw DSP code? Feb 04 12:31:10 according to warfare there's an ISP in Hamburg with a few maemo fans... in the crew Feb 04 12:31:13 you should be able to find a signature that indicates this kind of jump Feb 04 12:31:29 ?\ Feb 04 12:31:37 damo22: a jump from ARM to DSP? Feb 04 12:31:41 yeah Feb 04 12:31:49 that doesn't sound right at all Feb 04 12:32:13 well, at least in my case updating packages from the repos is still very slow Feb 04 12:32:24 or do they run independently? Feb 04 12:32:40 and what's worst, it seems to basically stall most of the time, jumping a few % from time to time Feb 04 12:32:43 sure they do Feb 04 12:33:05 as it seems to be working for others, I'm starting to wonder if I'm doing something wrong :) Feb 04 12:33:06 I found this with google Feb 04 12:33:07 http://img686.imageshack.us/img686/9440/rapuyamachipinfo.jpg Feb 04 12:33:32 jonwil: so I'm not sure if the contents are packed or they're just code for a different platform Feb 04 12:33:41 M4rtinK: thanks for reports. That's what we thought it is like, for a significant fraction of clients Feb 04 12:33:53 DocScrutinizer05: no problem Feb 04 12:33:57 it might be that the cmt firmware file is compressed Feb 04 12:34:07 oh and for the record, once I finally download a package, I get hash mismatch Feb 04 12:34:14 jonwil: it has a non-compressed header and non-compressed strings/ARM code in it Feb 04 12:34:20 hmmm ok Feb 04 12:34:35 basically always Feb 04 12:34:41 jonwil: yes, that's about all you ever get to know about rapuyama Feb 04 12:35:55 eddyb: is this the firmware for the phone part of the n900? Feb 04 12:36:02 That and information on how to remove the carrier lock from phones with BB5 chips Feb 04 12:36:04 damo22: yes Feb 04 12:36:22 eddyb: sounds like fun Feb 04 12:36:45 damo22: it's a secondary, less powerful SoC, but with DSP Feb 04 12:36:59 jonwil: yes, for all "normal" BB5 phones. unclear if it applies to N900 rapuyama Feb 04 12:37:27 not that it matters for N900 since very few, if any, were ever sold locked to a carrier Feb 04 12:37:40 :nod: Feb 04 12:37:44 there are some Feb 04 12:37:50 sprint iirc Feb 04 12:38:18 eddyb: maybe it will have special functions with low level calls to the hardware Feb 04 12:38:19 at least we got an unlock GUI Feb 04 12:38:51 probably from times when libisi didn't support USSD Feb 04 12:39:35 damo22: there's too little ARM code in there Feb 04 12:39:48 I dont think sprint ever sold N900s Feb 04 12:39:53 sprint is CDMA and the N900 is UMTS Feb 04 12:40:21 btw, I recon based on the evidence that the BB5 chip is made by Texas Instruments Feb 04 12:40:27 maybe the code is interleaved like a php file eg: htmlcraphere Feb 04 12:40:53 " Feb 04 12:41:31 bad example, since this is opcodes but still Feb 04 12:41:42 damo22: gaaah Feb 04 12:41:50 damo22: you have good ideas, but you're not helping Feb 04 12:41:53 me, I still see no point in bothering to reverse engineer the CMT firmware Feb 04 12:42:19 much better things to reverse engineer from where I sit :) Feb 04 12:42:52 ack Feb 04 12:43:13 I need to improve my analyzer to extract and disassemble the pieces of code Feb 04 12:43:15 eddyb: is there a pattern or consistent opcodes between the arm stuff and dsp stuff? Feb 04 12:43:54 damo22: oh, thanks, that reminded me, I wanted to look for patterns in the non-(ARM code/ASCII strings) parts of the binary Feb 04 12:44:33 :) Feb 04 12:44:59 If I had the time to really devote to it (and the right tools) I would go back and take another crack at the GPRS stuff... Feb 04 12:45:28 but thus far the connectivity UI bits have eluded pretty much all my attempts to reverse engineer them Feb 04 12:46:09 i am reverse engineering a sound card for firewire + alsa Feb 04 12:46:57 well, good luck Feb 04 12:47:05 playback is working Feb 04 12:49:19 but yeah hex-rays is great for x86, I use it all the time on any number of x86 binaries :) Feb 04 12:49:43 hmm, doesn't the SoC simply talk to whatever firmware runs on the BB5? Feb 04 12:50:15 jonwil: but no arm :( Feb 04 12:50:27 no, I haven't used hex-rays for ARM Feb 04 12:50:38 although I wish I had it to use on N900 stuffs :P Feb 04 12:50:50 what about objdump -c or whatever Feb 04 12:50:59 I wish I didn't have all this meaningless school stuff Feb 04 12:51:09 my small analyzer already beats IDA Feb 04 12:51:25 I have proof-of-return stack analysis Feb 04 12:51:41 Estel_: i suppose that we have to decide what the battery applet is supposed to be Feb 04 12:51:55 should it just report the data from hald-addon-bme and consider *that* as the most accurate? Feb 04 12:52:04 if so, the "problem" is with hald-addon-bme Feb 04 12:53:31 what is a good offline map solution for maemo? Feb 04 12:53:37 damo22: completely offline? Feb 04 12:53:43 well, yes, Feb 04 12:53:58 monav's offline routing is used pretty much by every open mapping solution Feb 04 12:54:00 it's *that* good Feb 04 12:54:01 except perhaps gps location currently displaying Feb 04 12:54:15 yeah, yeah, agps is cheap enough :) Feb 04 12:54:26 monav itself is... a bit weird, perhaps, but i like it Feb 04 12:54:30 and has vector maps Feb 04 12:54:37 i love vector maps Feb 04 12:54:39 modrana is also very good, but you have to download map tiles Feb 04 12:54:58 there's kde's marble, which is also kinda good Feb 04 12:55:00 does monav have whole world in vector format? Feb 04 12:55:14 I could have had a decompilator working months ago. or even just weeks. but no, I have to waste my time with this stupid school Feb 04 12:55:28 no you don't Feb 04 12:55:30 damo22: http://monav.openstreetmap.de/ these aren't *that* outdated Feb 04 12:55:53 damo22: the whole world is mapped, in separate packages Feb 04 12:55:57 it uses OSM data, of course Feb 04 12:56:01 kerio: wow nice Feb 04 12:56:22 afaik the version of monav in extras-devel doesn't work with those packages Feb 04 12:56:26 so you have to use the .deb from that site Feb 04 12:56:35 tadzik: talk to my parents, get me a place to stay and money for food Feb 04 12:56:43 ha, fair enough Feb 04 12:56:51 can i use the monav in stable and maps on the website? Feb 04 12:57:03 tadzik: after that, I *may* be able to code crazy awesome stuff for you Feb 04 12:57:16 damo22: i'm not sure if monav is in extras, just use the .deb from the website too :) Feb 04 12:57:23 I also could just be a dreamy teenager with a history of half-finished projects Feb 04 12:57:34 eddyb: that's... more likely Feb 04 12:57:41 kerio: thanks for the reality check Feb 04 12:58:03 who knows. my gf just dumped me. things are going up and up Feb 04 12:58:11 except I'm upside-down Feb 04 12:58:14 ...hand over your geek card Feb 04 12:58:25 eddyb: I thought so too. "When I finish my studies I'll have so much time!". Now I'm studying full-time and working half-time, and it's a bit exhausting sometimes Feb 04 12:58:47 real geeks dont have gfs? Feb 04 12:59:12 I'm fine being unreal then Feb 04 12:59:27 damo22: no, real geeks inject themselves with dopamine prepared in the basement lab Feb 04 12:59:37 you mean, pron Feb 04 13:00:11 take a look at this https://github.com/eddyb/wiREd/blob/master/disasm/x86.js#L167 Feb 04 13:00:23 isn't this better than any other disassembler out there? Feb 04 13:01:02 eddyb: HACK/FIXME/HACK/FIXME/FIXME in four consecutive lines Feb 04 13:01:03 nice Feb 04 13:01:36 I can be forgetful... I learned that the hard way (with dreams and ideas) Feb 04 13:02:00 I blame github for showing code before the highlighted line Feb 04 13:02:43 a disassembler written in javascript? Feb 04 13:03:07 why oh why Feb 04 13:03:15 damo22: stop whining, it's not PHP Feb 04 13:03:31 ES6 is the Perl of the decade Feb 04 13:03:42 whats wrong with C Feb 04 13:04:06 :P Feb 04 13:04:07 you're not going to get anywhere Feb 04 13:04:24 I can almost copy-paste the opcode tables from the intel manuals Feb 04 13:04:49 well, Perl and LISP Feb 04 13:04:49 (connection speed statistics) http://privatepaste.com/f0a869eb47 Feb 04 13:05:58 you could write a tokeniser in C that parses a binary and outputs a customised string for each byte on the input quite easily Feb 04 13:06:31 strings. pffft Feb 04 13:06:50 this baby doesn't output strings Feb 04 13:08:04 what does it output? the opcodes compiled into C? Feb 04 13:08:12 it outputs expressions Feb 04 13:08:32 similar to a simplified AST Feb 04 13:08:33 DocScrutinizer05: meh, good enough Feb 04 13:08:48 ok cool Feb 04 13:09:01 like chunks of assembler macros? Feb 04 13:09:45 damo22: like: EAX = 0; for xor eax, eax Feb 04 13:09:52 kerio: those are sorted top down top ~50 in download bandwidth Feb 04 13:10:06 kerio: lots of connections are at 0b 0b 0b Feb 04 13:10:07 damo22: or EIP = 0xf00bar; for jmp 0xf00bar Feb 04 13:10:10 DocScrutinizer05: oh :( Feb 04 13:10:39 DocScrutinizer05: can't we do throttling? Feb 04 13:11:52 lots of people are working on optimizing GCC to produce fast code, why dont they spend more effort writing a decompiler that outputs clean C code Feb 04 13:12:27 damo22: GCC is written in C. even if it's exaggerated, I consider myself a bit of a visionary Feb 04 13:12:40 you can't hold on to your old habits and have nice things at the same time Feb 04 13:12:50 progress is all about change Feb 04 13:13:15 kerio: [2013-02-04 13:27:04] warfare invents some conditional blacklisting based on criteria from /var/log/varnish/varnishncsa.log Feb 04 13:13:38 damo22, C isn't really that great a target language for a (de)compiler, you can easily encounter structures that don't nicely map into C. such as tail-jumps. Feb 04 13:13:46 (I'm to lame to code my ideas ;-P) Feb 04 13:14:10 what about an interpreted language like python Feb 04 13:14:27 flux: that too, but I think he was talking about the code the decompiler is written in Feb 04 13:14:38 damo22: python < JS < ES6 Feb 04 13:15:00 ? Feb 04 13:15:06 oh. I definitely wouldn't choose C for writing a disassembler :-o. I mean, what C's advantages you get to take advantage of? Feb 04 13:15:31 bison / flex or whatever Feb 04 13:15:37 pfffft Feb 04 13:15:43 right, those must be unique to C.. Feb 04 13:15:49 python > everything ?ú Feb 04 13:15:51 :) Feb 04 13:15:59 indeed Feb 04 13:16:10 M4rtinK: by that logic, whitespace > everything Feb 04 13:16:11 kerio: he just established that Feb 04 13:16:20 IIRC, it is used for RE quite often Feb 04 13:16:26 yeah Feb 04 13:16:33 but it's not that good Feb 04 13:16:35 python would be good i reckon Feb 04 13:16:44 gah please no more snake fans Feb 04 13:16:46 kerio: top ten from log get throttled to 75kbps now Feb 04 13:16:56 well, I've heard LOLCODE is quite nice :) Feb 04 13:17:17 well, everybody likes his own :) Feb 04 13:17:23 ok forget it, just fire up your DOS 6.22 with BASICA Feb 04 13:17:29 and I'm away for a session with my dentist :-S Feb 04 13:17:29 or QML ? :) Feb 04 13:17:45 there is QBS already - a build system Feb 04 13:17:51 based on QML Feb 04 13:17:53 a friend of mine is an admin at esolangs, he's fluent in ~20 programming languages Feb 04 13:18:01 and yet, his favourite is perl :s Feb 04 13:18:16 kerio: because it gets things done? Feb 04 13:18:30 just like Harmony Feb 04 13:18:34 and very well, I might add Feb 04 13:18:36 plenty of ways to get things done without having to use a write-only language Feb 04 13:18:47 ok i opened a can of worms Feb 04 13:18:51 sorry Feb 04 13:19:13 i think they all crawled away now though Feb 04 13:19:21 kerio: heh, Perl? I wouldn't learn it myself, at least nothing more than the basics Feb 04 13:20:02 but you cant blame me for being surprised you wrote a disassembler in js Feb 04 13:20:16 in a few years, legacy JS will be replaced, and all those problems you were whining about would be long gone Feb 04 13:20:25 uplink throttled from 25Mbps to 22 with that top ten punishment Feb 04 13:20:47 damo22: it's not technically JS, just the file extension. it's Harmony/ES6, "compiled" to js by google's traceur-compiler Feb 04 13:20:52 which also meant we shouldn't see packet loss on ISP side anymore Feb 04 13:20:58 means* Feb 04 13:21:13 DocScrutinizer05: 25MB/s? Or 25Mbit/s? Feb 04 13:21:17 k Feb 04 13:21:30 Mb/s Feb 04 13:21:41 That's pretty anemic. Feb 04 13:21:45 anaemic Feb 04 13:21:47 bit Feb 04 13:22:04 DocScrutinizer05: that's a bit too slow Feb 04 13:22:06 that's what ISP seems to shape our uplink to Feb 04 13:22:17 sth around 50..30 Feb 04 13:22:23 err 25..30 Feb 04 13:22:39 My $20/mo box sends out at 80-90Mbit/s no problem, and has been for the past 4-5 months or so. Feb 04 13:22:42 too slow for what? Feb 04 13:22:56 DocScrutinizer05: "bit". failed pun Feb 04 13:23:55 teotwaki: well, I don't know what contract Nemein has with their ISP Feb 04 13:24:06 hmm, seems to not to stall that much now Feb 04 13:24:33 M4rtinK: good :-) Feb 04 13:25:17 too soon - 74% and holding :) Feb 04 13:25:22 DocScrutinizer05: that being said, I was backing up rmo at noon... Feb 04 13:25:31 (from 296kB on the kmplayer package) Feb 04 13:26:44 how do battery applets on linux work? Feb 04 13:26:52 where do they get the data? Feb 04 13:27:02 depends Feb 04 13:27:11 sysfs, I think Feb 04 13:27:13 most of them use lshal or similar Feb 04 13:27:19 or a hal Feb 04 13:27:34 oh, acpi Feb 04 13:27:34 but the hal uses sysfs, right? Feb 04 13:27:49 no Feb 04 13:27:57 kerio: acpi on x86 only AFAIK Feb 04 13:28:06 hal uses haldaddonbme Feb 04 13:28:22 DocScrutinizer05: i meant on linux desktops Feb 04 13:28:46 meh Feb 04 13:28:48 Failed to fetch /kmplayer_0.10.6.900-3_armel.deb Hash Sum mismatch Feb 04 13:28:57 on desktops either hal or sysfs Feb 04 13:28:59 after it finished downloading Feb 04 13:29:00 M4rtinK: that's a repo problem Feb 04 13:29:21 that is, the download was probably successful Feb 04 13:30:32 toldya the hashsum is a problem inherited from original repo, introduced probably by autobuilder Feb 04 13:30:39 afaik Feb 04 13:31:24 I had no time to look into it and wrap my head around it Feb 04 13:31:40 ask thedead1440, merlin1991, brkn Feb 04 13:32:20 pester Nemein/x-fade to fix it Feb 04 13:42:27 M4rtinK: that hashsum error been at end of update? or did you do something else to trigger it? Feb 04 13:42:59 M4rtinK: like installing kmplayer from HAM Feb 04 13:44:16 logging on maemo.org not working Feb 04 13:44:25 invalid username or password Feb 04 13:44:34 odd, try to reset? Feb 04 13:46:07 nah, it's currently broken Feb 04 14:00:10 kerio: is there a way to display the street names on the offline vector maps in monav? Feb 04 14:00:19 i dunno Feb 04 14:00:54 thats the only thing i dont have atm Feb 04 14:01:40 damo22: directly in Monav ? Feb 04 14:01:47 M4rtinK: yeah Feb 04 14:02:03 damo22: depends if the renderer can do that Feb 04 14:02:24 and if street names are included in the vector data (they probably are) Feb 04 14:02:46 maybe try the desktop version if it shows up there ? Feb 04 14:03:19 DocScrutinizer05: I've just done "apt-get update" and then "apt-get install kmplayer" Feb 04 14:03:37 kmplayer because it showed there is an update for it Feb 04 14:03:55 I'll try to uninstall it first Feb 04 14:16:57 M4rtinK: I'm confused on it too; i just tried downloading kmplayer and it downloaded fine via apt-get install without any errors :S Feb 04 14:23:26 nope the vector maps dont display the street names in i386 either Feb 04 14:23:46 unless i go for online Feb 04 14:24:20 so it is probably renderer limitation then Feb 04 14:24:57 it is a shame Monav development is quite a bit in limbo at the moment Feb 04 14:25:50 which is unfortunate as many features are incomplete Feb 04 14:26:27 there are also quite a lot of clones with some interesting changes, but as the main developer seems to be AWOL, nobody is merging them back Feb 04 14:28:22 doh Feb 04 14:29:15 would the online maps use less data if i have the downloaded stuff? Feb 04 14:29:55 I think it is one or the other Feb 04 14:30:09 either you use the vector data for rendering Feb 04 14:30:20 or you use the prerendered map tiles Feb 04 14:30:33 I don't think it supports showing bot at the same time Feb 04 14:30:46 but I haven't played with it too much Feb 04 14:30:51 np Feb 04 14:31:06 (as I'm developing my own navigation system :) ) Feb 04 14:31:14 nice Feb 04 14:31:43 what about mapswithme Feb 04 14:31:56 is that maemo friendly? Feb 04 14:32:20 M4rtinK: gimme vector rendering! :( Feb 04 14:32:20 what is it ? Feb 04 14:32:26 downloading tiles is bullshit Feb 04 14:33:01 i had a iphone app that was free that used free vector maps, it rocked Feb 04 14:33:53 kerio: ok, if you help me resolve this: https://github.com/kothic/kothic/issues/5 :) Feb 04 14:34:30 once I can get Kothic to actually display something other than a blue screen (how fitting ! :) ) Feb 04 14:34:45 integrating it with the GTK interface should not be that difficult Feb 04 14:35:19 making a QPainter front end would be also nice :) Feb 04 14:41:19 thedead1440: trying again through SSH, so I can post the output Feb 04 14:44:40 for the record, the is the log from apt-get update: http://privatepaste.com/e5bee9f53b Feb 04 14:45:31 M4rtinK: so... Feb 04 14:45:43 open it in pdb, check what's options.filename? Feb 04 14:45:52 also, are you using py3k? Feb 04 14:46:48 kerio: I think that is is probably some projection error Feb 04 14:46:58 it does something with the vector data Feb 04 14:47:12 but probably for some offscreen area Feb 04 14:47:33 it is just that I haven't got to trying that yet Feb 04 14:47:55 so someone looking into it could speed up things quite a lot :) Feb 04 14:48:11 regarding Python 3 Feb 04 14:48:33 M4rtinK: just use python 2 Feb 04 14:48:43 I'm investigating how to write code that works both with Python 2.5 & 3.2 Feb 04 14:48:47 you can't Feb 04 14:48:50 they're wildly different languages Feb 04 14:48:54 and 2.5 is old as shit Feb 04 14:49:02 the only issues so far are string handling Feb 04 14:49:06 and exception handling Feb 04 14:49:13 those are both *HUGE* aspects Feb 04 14:49:19 well, not only but main Feb 04 14:49:34 2.5 is what N900 uses Feb 04 14:49:51 even the Neo FreeRunner is on 2.6 or 2.7 Feb 04 14:49:53 i am aware of that :) Feb 04 14:50:01 and meh, we could just upgrade Feb 04 14:50:20 we'll have to do that, at some point Feb 04 14:50:28 (to the latest 2.7) Feb 04 14:50:31 and Python 3.2 is what BB10 is using Feb 04 14:50:54 M4rtinK: just write your code in a way that 2to3 doesn't shit its pants Feb 04 14:51:03 yeah, Fremantle has the lowest Python version from any system I know modRana runs on :) Feb 04 14:51:33 having at least 2.6 would help due to all the aditional backports Feb 04 14:51:37 i wonder what i have that uses python Feb 04 14:52:05 M4rtinK: what's the issue there? All the messages are the warning for gpg keys that are expired on nokia's servers; nothing repo.maemo.org or mirrors can do but this shouldn't stop you from doin anything... Its a message everyone has been ignoring for months already Feb 04 14:52:06 oh, cleven and mobilehotspot Feb 04 14:52:13 ...i can live without either, really Feb 04 14:52:23 doing* Feb 04 14:52:32 thedead1440: yeah, it seemed nothing is wrong Feb 04 14:52:44 indeed nothing is wrong :) Feb 04 14:52:46 thedead1440: no, the problem is that right now, some repos on rmo are broken Feb 04 14:53:10 kerio: see his logs its just gpg key issue nothing else Feb 04 14:55:12 M4rtinK: just make modrana depend on python 2.7 :) Feb 04 14:55:40 kerio: as long there is none in Extras it is a no go Feb 04 14:55:56 I'm not a fan of dropping support for a platform just because I can :) Feb 04 14:56:02 http://maemo.org/packages/view/python2.7 Feb 04 14:56:46 kerio: so appo managed to get it in before everything broke down ? Feb 04 14:56:54 apparently Feb 04 14:57:03 ...but a -rc version, for some reason :s Feb 04 14:57:07 interesting Feb 04 14:57:35 the main questions are - how much space does it need Feb 04 14:57:55 - do PyGTK & PySide work with it ? Feb 04 15:00:03 thedead1440, DocScrutinizer05: so this is the full log I'm getting: http://privatepaste.com/65fd130f05 Feb 04 15:00:11 when trying to install kmplayer Feb 04 15:00:26 is probably the same of other packages Feb 04 15:00:59 M4rtinK: i don't think so Feb 04 15:02:23 this is what I get to install the python2.7 package, BTW: http://privatepaste.com/53c911615a Feb 04 15:24:44 Af'noon, al Feb 04 15:24:46 +l Feb 04 15:26:08 'lo Jaffa Feb 04 15:34:40 hi Feb 04 16:33:25 https://fbcdn-sphotos-c-a.akamaihd.net/hphotos-ak-prn1/48127_10151401467333908_1480734954_n.jpg Feb 04 16:37:14 RST38h, looks like hell. Feb 04 16:37:36 i am on archlinux, whenever i use flasher to flash my phone everything goes well for the first time but after that when i start flasher to flash again it cannot connect and says Suitable USB device not found, waiting. Feb 04 16:37:36 USB device found found at bus 001, device address 010. Feb 04 16:37:36 Error claiming USB interface: Device or resource busy Feb 04 16:38:09 General: Looks niiiice Feb 04 16:39:56 * RST38h has been there yesterday night, but that was before the snow storm. Looks much nicer tonight. Feb 04 16:41:59 timmy: have you blacklisted cdc_phonet? Feb 04 16:42:09 no Feb 04 16:42:14 what should i do? Feb 04 16:42:22 RST38h: quite a little snow on the photo :) Feb 04 16:43:24 Well, they are constantly clearing it there. Feb 04 16:43:28 timmy: hold on, you said that the first flash goes well? Feb 04 16:43:34 is the first flash VANILLA? Feb 04 16:44:00 first one goes well but it can be anything. Feb 04 16:44:11 even flashing a kernel Feb 04 16:44:26 hm, weird Feb 04 16:44:30 i had no such problem on debian, but here on archlinux i am facing this problem Feb 04 16:44:32 are you rebooting the n900? Feb 04 16:44:42 also, lsmod | grep cdc Feb 04 16:45:30 cdc_phonet 4550 0 Feb 04 16:45:30 phonet 20566 1 cdc_phonet Feb 04 16:45:30 usbcore 148406 7 btusb,uhci_hcd,uvcvideo,ums_realtek,usb_storage,ehci_hcd,cdc_phonet Feb 04 16:45:36 fukken knew it Feb 04 16:45:55 disconnect the n900, modprobe -r cdc_phonet Feb 04 16:45:55 what is it? Feb 04 16:45:57 and then blacklist it Feb 04 16:46:11 ok Feb 04 16:46:12 tnx Feb 04 18:00:37 curious for those long time n900 users, ever had one die for no apparent reason (versus mechanical/physical failure or user-caused failures?) Feb 04 18:02:35 sure Feb 04 18:03:00 just wondering about construction/chip quality... Feb 04 18:03:18 and how long to expect this thing to last if taken care of, physically... Feb 04 18:03:52 5+ years Feb 04 18:03:58 of constant on Feb 04 18:04:17 depends on your luck though Feb 04 18:04:20 cool. Feb 04 18:05:12 modem is known to fail for no obvious reason, for some users Feb 04 18:05:36 probably soldering issues or actually chip wear Feb 04 18:15:42 ugh. Feb 04 18:16:35 modem as in, the GSM module, so it can't make call anymore either? Feb 04 18:16:38 yep Feb 04 18:16:44 gps will also be busted Feb 04 18:17:23 gps is slow to make locks to begin with... Feb 04 18:17:32 use supl.google.com Feb 04 18:17:37 my 5230 locked on GPS much faster Feb 04 18:17:42 (also AGPS) Feb 04 18:17:50 i doubt your 5230 locked faster than 4 seconds Feb 04 18:17:56 at least, not dramatically faster Feb 04 18:18:02 well, it was faster than the n900 Feb 04 18:18:22 yeah Feb 04 18:18:33 the N900 is one of the most powerful phones out there Feb 04 18:18:35 but also the slowest. Feb 04 18:18:43 every other Nokia was faster Feb 04 18:18:51 the 5230's cpu is weak though :D Feb 04 18:19:19 I think it has a lot to do with the full multitasking of Maemo Feb 04 18:19:40 am i a bad enough dude to install g++, i wonder Feb 04 18:19:41 no wonder why every other smartphone OS doesn't have full multitasking ... Feb 04 18:19:50 I'm impressed by maemo multitasking. Feb 04 18:20:01 or, as it's known in other circles, "multitasking" Feb 04 18:20:27 like, the one that you'd get from a preemptive OS Feb 04 18:20:41 the 5230 symbian multitasking is fail. Feb 04 18:20:54 it's enough to get around though. Feb 04 18:21:03 the symbian multitasking was perfect, for its time Feb 04 18:21:20 I absolutely despised the fact that it dropped gsm link when you quit out of the browser Feb 04 18:21:38 the PalmOS "multitasking" was also great Feb 04 18:21:44 that's not a multitasking problem Feb 04 18:21:53 that's a connection ui problem Feb 04 18:22:04 that has never happened to me Feb 04 18:22:11 it should have multiplexed the network correctly... Feb 04 18:22:13 Sc0rpius: yes it did Feb 04 18:22:33 it's not that it doesn't multiplex it, it's just that it killed connections when there was no program using them Feb 04 18:22:34 my 3G connection is very stable in the N900 Feb 04 18:22:41 and I'm opening/closing the browser all the time Feb 04 18:22:47 no, we're talking about symbian here Feb 04 18:22:51 ooohh Feb 04 18:23:00 I loved the way symbian managed connections Feb 04 18:23:13 I always knew if a program was transmitting or not Feb 04 18:23:35 and you could connect browser over 3g the same time e.g. torrent was downloading over wifi Feb 04 18:23:37 awesome Feb 04 18:23:45 it mostly worked though. Except for those connection issues. Feb 04 18:24:12 I wonder if the Ubuntu phones will have full multitasking like Maemo, that is (for purists) if it's gonna be preemptuve Feb 04 18:24:15 preemptive Feb 04 18:24:19 with its scheduler and everything. Feb 04 18:24:31 linux is preemptive... it's all presentation... Feb 04 18:24:38 This was very well thought regarding expensive 3g. On N900 ham might kill your quota in minutes Feb 04 18:25:04 yeah but "Ubuntu phone" is not a full Linux kernel anyway so we have to worry what will be stripped out of it Feb 04 18:26:08 Sc0rpius, no? and how will desktop mode work without full kernel? Feb 04 18:26:51 you mean the "superphone" ? I wonder if that's gonna be real. Feb 04 18:27:00 because the battery will last like 10 minutes Feb 04 18:27:09 stripped out? Feb 04 18:27:53 I've never actually used an iphone(or even android even)... not sure how task switching works on them... but the mini windows on the task selector in maemo is impressive, I thought they'd be static images... Feb 04 18:28:07 there are no task switching Feb 04 18:28:17 the application that's running on the frontend is the only application running Feb 04 18:28:24 well, there is "task switching" Feb 04 18:28:27 and apps register some "events handlers" Feb 04 18:28:46 you can switch the current application to another one Feb 04 18:28:49 obviously Feb 04 18:28:59 but there's always just one app running nothing else Feb 04 18:29:23 if it's possible for the foreground application to completely take over the whole touchscreen, that's a problem... Feb 04 18:30:18 that's the case for my 5230 for the native icq client, but at least the hardware buttons could break out of the application) Feb 04 18:30:45 eccerr0r: there's little to no task switching on iOS Feb 04 18:30:51 for a precise reason Feb 04 18:31:10 and applications are always fullscreen, in some way Feb 04 18:32:59 kerio: the specs war has made even iOS capable to mult-task in different ways... see: https://www.youtube.com/watch?v=5Q_DvfOOeJE Feb 04 18:34:04 hey, it's still a bsd kernel underneath Feb 04 18:34:49 indeed but due to the specs war things which were once unthinkable are now doable with work-arounds on iOS too Feb 04 18:39:21 do you guys reckon that /usr/include can be optified? Feb 04 18:40:01 well in the iOS case no task switching is "great" Feb 04 18:40:11 I have an iPad 2 that I use a lot and the battery last AGES Feb 04 18:40:14 AAAAAAAAAAAAAGEEEEEEEEES Feb 04 18:40:17 and that's great though Feb 04 18:40:26 I would sacrifice multitasking for that in a phone Feb 04 18:41:02 my N900 lasts like 4 hours in 3G Feb 04 18:42:55 Sc0rpius: my colleague tested linux on Apple HW and said it's 6h on batteries for iOS versus 1,5 hour for linux Feb 04 18:43:54 ok, optified include Feb 04 18:43:57 i hope everything goes well Feb 04 18:46:12 guess who's top bandwidth hog on our rmo Feb 04 18:46:28 121.203.146.188.in-addr.arpa. 43200 IN PTR 188.146.203.121.nat.umts.dynamic.t-mobile.pl. Feb 04 18:46:35 :o Feb 04 18:48:20 are the repositories back? Feb 04 18:48:27 which also means our throttling of top10 non-APT-maemo clients to 75kBit/s works just fine Feb 04 18:48:36 cehteh: basically yes Feb 04 18:48:57 \o/ Feb 04 18:49:10 DocScrutinizer05: i had a less-than-10kb/s download to prove you wrong Feb 04 18:49:23 cehteh: we're just facing quite a few bandwidth problems Feb 04 18:49:39 * cehteh just hits 'update' Feb 04 18:49:56 lucky you that i dont have 1k devices :P Feb 04 18:49:57 kerio: you prove me wrong? on what, Sir? Feb 04 18:50:08 on the fact that it doesn't work just fine! >:c Feb 04 18:50:17 BS Feb 04 18:50:29 looks stuck here Feb 04 18:50:48 the fact that you didn't get >75kBit/s proves it's working excellent Feb 04 18:51:19 will there be a round robin mirrored r.m.o sometime? Feb 04 18:51:34 sometime, maybe Feb 04 18:51:40 DocScrutinizer05: let's shutdown the r.m.o. and nobody can get >75kBit/s ;) Feb 04 18:51:49 Gentoo portage would suck if it were all dumped ont one server... Feb 04 18:51:57 luf: hahaha Feb 04 18:52:57 DocScrutinizer05: let's try to count how long it takes to download 9MB (extras-dev) abd 3,5MB (testing or tools) with 10kBit/s speed ... Feb 04 18:53:03 is there an etiquette of updates, like Gentoo tries to stop people from syncing more than once a day... Feb 04 18:53:11 luf: I could put 89.102.208.114 on the blacklist manually Feb 04 18:53:48 luf: why do you think I'm interested in such silly math games? Feb 04 18:53:56 isn't that for kindergarden? Feb 04 18:53:58 DocScrutinizer05: fell free to do it. My wife will appreciate it. Feb 04 18:54:55 honestly, I didn't come here to get flames for trying to manage available bandwidth for you >:-( Feb 04 18:54:57 DocScrutinizer05: Maybe I red it wrong but it seems to me you're satisfied with the current situation. Feb 04 18:55:13 read again! Feb 04 18:56:08 cehteh: just another 1k? pfff Feb 04 18:56:11 is it mostly the repository indexes that's being hammered or actual package files? Feb 04 18:56:24 packages.gz Feb 04 18:56:33 so the index. Feb 04 18:56:43 kerio, I don't get the dillema Feb 04 18:56:44 DocScrutinizer05: I'm sorry I'm too tired today. Feb 04 18:56:49 and suckers running wget against whole repo, which we throttled down to 75kBit/s now Feb 04 18:56:54 if bq27x00 is calibrated, use it for everything Feb 04 18:56:59 ~mirrors Feb 04 18:57:00 somebody said mirror was http://maemo-archive.wedrop.it/ http://maemo.merlin1991.at/apt-mirror/ http://talk.maemo.org/showthread.php?p=1315143#post1315143 Feb 04 18:57:01 bars, percents, capacity, everything Feb 04 18:57:27 If not, use capacity, bars, etc from hald-addon-bme (but current and voltage still from bq27x00) Feb 04 18:57:44 if bq27x00 isn't available at all (no module loaded) use hald-addon-bme Feb 04 18:57:48 problem solved Feb 04 18:57:49 Estel_: my question is different Feb 04 18:58:13 now, it uses prorities properly, but still take bars from rx51 or god knows what Feb 04 18:58:20 Estel_: ack Feb 04 18:58:27 = it sayd 47% but shows full bars on battery Feb 04 18:58:32 Estel_: install +kerio1 Feb 04 18:58:35 DocScrutinizer05, ? Feb 04 18:58:39 but that's not my question, really Feb 04 18:58:41 kerio, I know, thanks Feb 04 18:58:47 Estel_: you're right Feb 04 18:58:51 but why you have disabled everything except bq27x00 Feb 04 18:58:56 DocScrutinizer05, ah, thanks Feb 04 18:59:16 my question is, should hald-addon-bme report the data from bq27k? it's supposed to be the "official" way to get battery data in maemo Feb 04 18:59:21 together with libbmeipc Feb 04 18:59:31 and how it gets data now? Feb 04 18:59:36 if so, then the battery applet becomes just a frontend for hald-addon-bme, and it makes more sense Feb 04 18:59:37 I need to get a copy of the reflash images someday. seems tablet-dev.nokia.com has died too :( Feb 04 18:59:56 Estel_: from the modules, but stuff like "design" is taken from rx51, for instance Feb 04 19:00:03 kerio, ok, it's dillema for future, but now, why won't we set applet to do priorities as I just said Feb 04 19:00:28 kerio: fsck "design", make that 1320 hardcoded! Feb 04 19:01:05 also, pali seems fixated in wanting to use design when he should be using full_charge Feb 04 19:01:24 well, as said bq27x00 when available, if uncalibrated, take capacity and % from whatever else (hald-addon-bme from rx51), and if no module at all, take everything from hald-addon-bme (rx51) Feb 04 19:01:29 maybe with a way to override the defualt on module load time, with a module parameter Feb 04 19:01:54 DocScrutinizer05, won't happen, Pali doesn't want to use anything non-upstreamed on bq27x00 Feb 04 19:02:00 so no custom Maemo settings Feb 04 19:02:10 pfff Feb 04 19:02:17 DocScrutinizer05: indeed ._. Feb 04 19:02:26 unless someone make patch and upstream it - to make parameter at load time accepted - no way :/ Feb 04 19:02:27 this shit is FULL OFF maemo settings Feb 04 19:02:28 anyway Feb 04 19:02:34 DocScrutinizer05, but upstreamed Feb 04 19:02:38 starting with 20mR value of RS Feb 04 19:02:56 he just doesn't want to make maintaining harder Feb 04 19:03:27 which is quite understandable. Well, seriously though, just screw it and do as proposed, bq27x00 when available and calibrated, and everything else if not Feb 04 19:03:43 how t f will maintaining get harder when you assume sane defaults and implement a standard way to override them? Feb 04 19:04:11 particularly when compared to the alternative that seems like using random generator for "design capacity" Feb 04 19:04:14 I would add middle-variant of bq27x00 available but not calibrated - in such case, everything *ėxcept* capacities and % would be from bq27x00 too, - but i can live without this variant, too Feb 04 19:04:19 Estel_: my issue is that we should, instead, get hald-addon-bme to report the correct values instead Feb 04 19:04:25 in hal-device bme Feb 04 19:04:44 so the battery applet becomes everything-agnostic Feb 04 19:04:52 kerio, I have no opinion here, as long as end-user result is what I said - bq27x00 > rx51_battery Feb 04 19:05:10 I don't care how it's processed internally, frankly Feb 04 19:05:31 I just want this applet to consider 3 simple variants, no matter if it's frontend, half backend, or whatsnot Feb 04 19:05:45 afaik nothing else cares about capacity Feb 04 19:05:55 reporting correct values in /org/freedesktop/Hal/devices/bme is better Feb 04 19:06:00 well, BTW, why not make hald-addon-bme follow same rules? Feb 04 19:06:01 ok Feb 04 19:06:02 so Feb 04 19:06:10 and also making the battery applet use last_full instead of design Feb 04 19:06:17 if bq27x00 is available pand calibrated, hald-addon-bme takes everything from it Feb 04 19:06:19 because why the hell should i care of how full the battery is compared to when it was built? Feb 04 19:06:24 sure Feb 04 19:06:27 so, as i said. Feb 04 19:06:33 yes, as you said Feb 04 19:06:34 :) Feb 04 19:06:41 we got hald-addon-bme replacement that should hide any hw side differneces from hal and thus from battery applet Feb 04 19:06:54 if bq27x00 not calibrated but available, take everything from bq27x00 too, but capacity and percent from rx51_battery Feb 04 19:07:05 DocScrutinizer05: indeed, but it's not what the replacement battery applet does, for compatibility with hybrid situations Feb 04 19:07:09 if no bq27x00 at all, hald-addon-bme takes everything from rx51_battery Feb 04 19:07:11 problem solved Feb 04 19:07:19 what does rx51_battery provide, anyway? Feb 04 19:07:26 have fun with that, guys Feb 04 19:07:35 are there mirros for the reflash images and flasher? or do we have to wait till nokia fixes their server? Feb 04 19:07:57 ~mirrors Feb 04 19:07:58 it has been said that mirror is http://maemo-archive.wedrop.it/ http://maemo.merlin1991.at/apt-mirror/ http://talk.maemo.org/showthread.php?p=1315143#post1315143 Feb 04 19:08:00 I think in long run we don't need to care about hybrids, as bq27x00_battery ism part of replacement, as much as new hald-addon-bme Feb 04 19:08:07 Estel_: the problem is that pali will start going "i want the battery applet to work even when users have stopped bme and loaded the kernel modules", except in a more broken english Feb 04 19:08:13 but, for now, some hybrid approach may be feasible Feb 04 19:08:21 ~skeiron Feb 04 19:08:26 eccerr0r: skeiron.org/tablets-dev Feb 04 19:08:32 kerio, nope, as user is never meant to stop replacement bme Feb 04 19:08:43 as replacement bme doesn't interfere with anything Feb 04 19:08:53 coff coff i2cget coff coff Feb 04 19:08:59 You MUST NOT stop replacement bme Feb 04 19:09:02 ahh, ok. Feb 04 19:09:13 what do you need i2c access with replacement bme for? Feb 04 19:09:15 eccerr0r: i'm not sure about the long-term solution for that Feb 04 19:09:56 I just released new version of my BNF scriptset, that replaces everything that relied on i2c, for relying on bq27x00 (and failsback to i2c if no module). If I was able to do it overnight, others can too Feb 04 19:10:13 do we need i2c for anything else? Feb 04 19:10:14 ~skeiron is hte semi-official backup and emergency standin for all internet borne maemo resources: http://http://skeiron.org/tablets-dev/ http://talk.maemo.org/showthread.php?p=1315143#post1315143 Feb 04 19:10:14 DocScrutinizer05: okay Feb 04 19:10:19 need to set up a bittorrent :D Feb 04 19:10:43 infobot: no, skeiron is the semi-official backup and emergency standin for all internet borne maemo resources: http://http://skeiron.org/tablets-dev/ http://talk.maemo.org/showthread.php?p=1315143#post1315143 Feb 04 19:10:43 DocScrutinizer05: okay Feb 04 19:11:20 DocScrutinizer05: no fremantle/tools Feb 04 19:11:21 how loaded down is skeiron, or is there a time of day that would be better for me to download it? Feb 04 19:11:59 oh, there is, but the bzip2ped Packages is broken Feb 04 19:12:02 kerio, DocScrutinizer05, any other practical uses for i2c other than battery chip info? Feb 04 19:12:05 kerio, anyway Feb 04 19:12:35 if no bme available, but module is loaded, use it. If module available but not calibrated and no bme, use this damn design value and show middle finger Feb 04 19:12:39 Estel_: there are like several dozen subsystems on 3 I2C busses, on N900 Feb 04 19:12:41 problem solved anyway Feb 04 19:12:46 just two more IF's Feb 04 19:12:49 DocScrutinizer05, I know Feb 04 19:13:05 eccerr0r: don't worry Feb 04 19:13:11 I just wonde if we use i2c raw access for any practical purpose except battery chip Feb 04 19:13:14 use it and profit Feb 04 19:13:30 who's "we"? Feb 04 19:13:30 okay. since these are fairly large files :D Feb 04 19:13:49 DocScrutinizer05, we - "family" of maemo related people Feb 04 19:14:00 i.e. anyone ever used it for anything else and lived to tell the tale Feb 04 19:14:07 and produce something productive using it Feb 04 19:14:17 no, "we" usually don't care about I2C, there are kernel drivers doing that for us Feb 04 19:14:29 thats what I thought Feb 04 19:14:30 anyway Feb 04 19:14:36 if someone want raw i2c access Feb 04 19:14:52 he can disable bme and live with innacurate battery capacity readings for fckin 5 mins Feb 04 19:14:56 methinks Feb 04 19:15:15 yup, that's the general tale Feb 04 19:15:41 if you want access an I2C chip directly, you first need to unload the kernel module driver Feb 04 19:16:16 anyway, we should convince pali to provide the most correct data in hald-addon-bme Feb 04 19:16:47 kerio, to convince Pali to anything, You MUST create a wiki page Feb 04 19:16:51 ~must Feb 04 19:16:59 ~musn't Feb 04 19:17:04 ~useless Feb 04 19:17:05 * infobot starts crying and hides from estel_ in the darkest corner of the room. :( Feb 04 19:17:29 as til tommorow no one will remember simple solutions from today, posted on irc Feb 04 19:18:30 btw that pun over Pali's english wasn't nice, kerio Feb 04 19:18:41 reminds me of tetowaki's bitching yesterday Feb 04 19:19:18 ~2119 Feb 04 19:19:19 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119. Feb 04 19:19:46 Estel_: i feel somewhat authorized to mock him, i'm not a native english speaker either Feb 04 19:20:05 ohp thanks Feb 04 19:20:14 and it shows, sometimes, but i put the effort in at least :s Feb 04 19:20:27 ~#maemo must is also ~2119 Feb 04 19:20:28 okay, Estel_ Feb 04 19:20:34 ~#maemo musn't is also ~2119 Feb 04 19:20:35 okay, Estel_ Feb 04 19:20:40 doesn't work that way i think Feb 04 19:20:44 ~must Feb 04 19:20:44 extra, extra, read all about it, must is ~2119 Feb 04 19:20:56 infobot: no, must is see 2119 Feb 04 19:20:57 it was what I meant Feb 04 19:20:57 kerio: okay Feb 04 19:21:00 ah Feb 04 19:21:03 even better Feb 04 19:21:05 infobot: no, mustn't is see 2119 Feb 04 19:21:06 kerio: okay Feb 04 19:21:09 ~must Feb 04 19:21:10 well, must is ~2119 Feb 04 19:21:15 :P Feb 04 19:21:15 ah crap Feb 04 19:21:32 DocScrutinizer05: i screwed up default must in infobot Feb 04 19:21:36 is there a way to recover it? Feb 04 19:21:53 infobot, no, must is The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119. Feb 04 19:21:54 Estel_: okay Feb 04 19:21:57 ~must Feb 04 19:21:58 must is, like, ~2119 Feb 04 19:22:02 infobot: no, #maemo must is see 2119 Feb 04 19:22:03 okay, kerio Feb 04 19:22:06 infobot: forget about must Feb 04 19:22:07 kerio: i didn't have anything called 'about must' to forget Feb 04 19:22:12 infobot: forget must Feb 04 19:22:12 ~forget must Feb 04 19:22:13 i forgot must, kerio Feb 04 19:22:16 infobot: forget mustn't Feb 04 19:22:17 i forgot mustn't, kerio Feb 04 19:22:27 infobot: #maemo mustn't is see 2119 Feb 04 19:22:28 okay, kerio Feb 04 19:22:30 ~must Feb 04 19:22:31 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119. Feb 04 19:22:33 there Feb 04 19:22:44 this was harder than it should've been Feb 04 19:22:51 ~musn't Feb 04 19:22:51 somebody said musn't was ~2119 Feb 04 19:23:02 ~mustn't Feb 04 19:23:03 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119. Feb 04 19:23:05 screwed Feb 04 19:23:06 infobot: forget musn't Feb 04 19:23:07 i didn't have anything called 'musn't' to forget, kerio Feb 04 19:23:13 infobot: forget #maemo musn't Feb 04 19:23:14 kerio: i forgot #maemo musn't Feb 04 19:23:15 ~forget mustn't Feb 04 19:23:16 Estel_: i didn't have anything called 'mustn't' to forget Feb 04 19:23:29 ~must Feb 04 19:23:30 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119. Feb 04 19:23:33 ~musn't Feb 04 19:24:27 infobot: #maemo musn't is see 2119 Feb 04 19:24:28 okay, Estel_ Feb 04 19:24:33 ~musn't Feb 04 19:24:34 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119. Feb 04 19:24:39 wtf is "musn't" Feb 04 19:24:45 must not? Feb 04 19:24:50 nope Feb 04 19:25:08 so blame my english too ;) Feb 04 19:25:16 infobot: forget #maemo musn't Feb 04 19:25:16 i forgot #maemo musn't, kerio Feb 04 19:25:35 it should be mustn't Feb 04 19:25:45 yea Feb 04 19:25:55 even though it is pronounced musn't but it is mustn't Feb 04 19:25:55 damn, it's harder to write than must not Feb 04 19:26:03 yeah hehe Feb 04 19:26:13 yes yes, got my brainwave activated again, about it Feb 04 19:26:23 thanks anyway Feb 04 19:32:27 don't mess with infobot factoids when you don't know what you're doing Feb 04 19:32:53 ~forget _default mustn't Feb 04 19:32:54 DocScrutinizer05: i didn't have anything called '_default mustn't' to forget Feb 04 19:33:31 ~unforget _default mustn't Feb 04 19:33:31 DocScrutinizer05: that factoid was not backedup :/ Feb 04 19:33:40 meh Feb 04 19:35:02 ~forget _default musn't Feb 04 19:35:02 i didn't have anything called '_default musn't' to forget, DocScrutinizer05 Feb 04 19:35:19 we cleaned it up, afterwards. Feb 04 19:35:49 anyway, kerio, you ~must create wiki page to convince Pali about anything... Installing your fixed deb, thanks! Feb 04 19:38:56 DocScrutinizer05: i was confused by the channel prefix :( Feb 04 19:39:03 anyway, any generic factoid for "must" is clearly stupid Feb 04 19:54:51 kerio: yep, that can get confusing sometimes. Zhe more important is to check a factoid with ~literal before redefining it Feb 04 20:05:47 kerio: however, I must say I don't see how a #maemo specific refernece to RFC2119 makes more sense than a _default factoid for 'must' (if there ever been any) Feb 04 20:34:44 kerio: ping Feb 04 20:34:49 pong Feb 04 20:35:01 ~ding Feb 04 20:35:01 dong Feb 04 20:35:20 what do I have to put on 3.8rc3 cmdline? Feb 04 20:35:25 or it was Pali Feb 04 20:35:27 ? Feb 04 20:41:10 kerio: it is Pali Feb 04 20:41:15 Pali ping Feb 04 21:12:31 Anyone able to download shortcutd atm? Fails when I try to install via HAM Feb 04 21:17:26 n900-dk: are you trying to use rmo? Feb 04 21:18:17 yeah, it downloads but gives 'Hash Sum mismatch' Feb 04 21:18:33 n900-dk: from which repo? -devel? Feb 04 21:18:37 use extras-devel.merlin1991.at Feb 04 21:18:42 ask somebody else, then - I cannot even connect at the moment Feb 04 21:19:23 n900-dk: I'm also getting hash mismatch Feb 04 21:19:31 kerio: rmo - extras-devel Feb 04 21:19:36 on every package I've tried so far Feb 04 21:19:39 yeah, use merlin1991's mirror Feb 04 21:19:41 is the bestest Feb 04 21:19:49 kerio: thx Feb 04 21:20:10 well, I would rather like to know what's wrong :) Feb 04 21:20:33 maybe it gets corrupted during to the excruciatingly slow download ? :) Feb 04 21:22:01 nah, it's just that the last repo update was interrupted somehow Feb 04 21:22:06 possibly due to the migration Feb 04 21:23:32 M4rtinK: it's most probably a bug in autobuilder Feb 04 21:23:54 merlin1991: ping Feb 04 21:24:46 merlin1991: you looked into that, together with thedead1440, no? Could you enlighten us please what's been your findings and conclusions Feb 04 21:25:23 Findings: "It doesn't work". Conclusion: "It still doesn't work". Feb 04 21:25:25 so recently built packages might be corrupted ? Feb 04 21:25:45 I'll try something older Feb 04 21:25:57 teotwaki: not helpful Feb 04 21:26:14 DocScrutinizer05: it's jolly banter, it's not supposed to be. Feb 04 21:26:57 How much bandwidth do we have for rmo? Feb 04 21:27:23 n900-dk: not enough Feb 04 21:27:55 kerio: :) Feb 04 21:28:50 DocScrutinizer05: I'm trying libavutil49 Feb 04 21:29:14 last update is from 2011, so it should be free from any autobuilder induced corruption Feb 04 21:29:42 DocScrutinizer05: installed fine Feb 04 21:31:38 Oh, seems like it's 25mbps - wonder what the bandwidth was before. 25mbps is also quite a lot for not so many n900s Feb 04 21:33:25 it seems to be maxed most of the time Feb 04 21:33:44 BTW, the akamai statistics showed about 40k N900s Feb 04 21:34:49 30k Feb 04 21:34:57 40k maemo devices Feb 04 21:35:24 oh yeah ale all the N<900 Feb 04 21:36:31 DocScrutinizer05: so there's still 10k n810s? Feb 04 21:36:36 but still, as long as they don't attack rmo at the same time, I guess 25mbps could be enough Feb 04 21:37:13 it's not Feb 04 21:37:37 40k = 0.5 updates / second Feb 04 21:38:12 0.5 update = 3, 5 or 8 MB / 2 Feb 04 21:38:59 4MB /s = 40Mbit/s Feb 04 21:39:09 plus we got other leechers too Feb 04 21:39:25 though we throttled the 10 worst of them Feb 04 21:39:27 4MB/s = 32Mbit/s Feb 04 21:39:45 ok, ok, I'll stfu. Feb 04 21:40:04 better, before you lose Feb 04 21:40:18 when I start about overhead Feb 04 21:40:39 aye Feb 04 21:40:48 the TCP overhead on small files is ridiculous Feb 04 21:41:03 FTP is even worse. Feb 04 21:41:16 this is HTTP afaik Feb 04 21:42:01 aye, which is why I said "FTP is even worse" Feb 04 21:42:21 * ShadowJK would guess 1-2k overhead Feb 04 21:42:22 Could be interesting to see the MRTGs Feb 04 21:42:47 MRTGs are roughly useless on file servers. Feb 04 21:43:01 top and iftop, and sar, are the only useful tools. Feb 04 21:43:14 iftop Feb 04 21:43:19 is what I use Feb 04 21:43:33 DocScrutinizer05: what's the iowait on the server? Feb 04 21:44:07 top is not interesting any more, since varnish dropped CPU load to 98% idle Feb 04 21:44:28 top; s0.1 Feb 04 21:44:36 top - 21:43:37 up 11 days, 3:46, 1 user, load average: 0.07, 0.11, 0.14 Feb 04 21:44:37 Tasks: 110 total, 1 running, 109 sleeping, 0 stopped, 0 zombie Feb 04 21:44:39 Cpu(s): 0.1%us, 1.0%sy, 0.0%ni, 91.9%id, 4.7%wa, 0.0%hi, 1.0%si, 1.3%st Feb 04 21:44:40 Mem: 2049296k total, 2032840k used, 16456k free, 44388k buffers Feb 04 21:44:42 Swap: 2097144k total, 17860k used, 2079284k free, 1284840k cached Feb 04 21:44:50 so still 5% iowait Feb 04 21:45:05 only GB ram? Feb 04 21:45:16 so the 25mbps are maxed all day? Feb 04 21:45:17 s/y/y 2/ Feb 04 21:45:18 teotwaki meant: only 2 GB ram? Feb 04 21:45:24 yes, all day Feb 04 21:45:28 and night Feb 04 21:45:35 only 1 user? Feb 04 21:45:42 that is very uncommon Feb 04 21:45:43 WizardNumberNext: one tty Feb 04 21:46:03 WizardNumberNext: it would be uncommon to see more than 3 on a server. Feb 04 21:46:11 indeed Feb 04 21:46:21 I see, I got it wrong Feb 04 21:46:49 How much extra will they charge us for 50mbps? Feb 04 21:46:53 top tells me 20 users, but I have multiple sessions of konsole all around 9 dekstops Feb 04 21:46:58 http://talk.maemo.org/showthread.php?p=1320780#post1320780 Feb 04 21:47:23 but 4.7 wait is really a lot Feb 04 21:47:27 WizardNumberNext: yeah, even for a desktop that's a lot Feb 04 21:47:45 WizardNumberNext: not quite, want to see the output of one of my servers? Feb 04 21:48:11 wait, what's "wait" in cpu measurement? Feb 04 21:48:27 teotwaki are you using i386 to get more then that? Feb 04 21:48:43 wait is waiting for pretty much everything Feb 04 21:48:50 oh, they will add it for free? Or just temporary? Feb 04 21:48:52 DocScrutinizer05: Thanks! Feb 04 21:48:58 why would you use cpu to wait? :o Feb 04 21:49:06 top - 22:48:59 up 75 days, 5:24, 1 user, load average: 1.23, 1.20, 1.20 Feb 04 21:49:06 Tasks: 105 total, 2 running, 103 sleeping, 0 stopped, 0 zombie Feb 04 21:49:06 Cpu0 : 16.3%us, 2.7%sy, 0.0%ni, 6.6%id, 72.4%wa, 0.0%hi, 0.0%si, 2.0%st Feb 04 21:49:06 Cpu1 : 0.3%us, 0.3%sy, 0.0%ni, 99.3%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st Feb 04 21:49:06 Mem: 4194304k total, 4183076k used, 11228k free, 96812k buffers Feb 04 21:49:07 Swap: 2097144k total, 92k used, 2097052k free, 3332500k cached Feb 04 21:49:28 because not everything can be written to use interrupt in common-sense Feb 04 21:49:29 kerio; waiting for disk I/O Feb 04 21:49:41 disk or network Feb 04 21:49:50 in a busyloop? Feb 04 21:50:35 top - 21:47:56 up 14 days, 8:14, 2 users, load average: 0.22, 0.15, 0.11 Feb 04 21:50:36 in this case, this server streams binary logs from a fileserver, parses them and inserts the lot of data in a mysql server Feb 04 21:50:36 Tasks: 359 total, 1 running, 358 sleeping, 0 stopped, 0 zombie Feb 04 21:50:38 Cpu0 : 0.1%us, 0.2%sy, 0.1%ni, 99.4%id, 0.1%wa, 0.0%hi, 0.2%si, 0.0%st Feb 04 21:50:39 Cpu1 : 0.1%us, 0.2%sy, 0.1%ni, 99.3%id, 0.3%wa, 0.0%hi, 0.0%si, 0.0%st Feb 04 21:50:41 Cpu2 : 0.1%us, 0.2%sy, 0.1%ni, 99.5%id, 0.1%wa, 0.0%hi, 0.0%si, 0.0%st Feb 04 21:50:42 Cpu3 : 0.1%us, 0.2%sy, 0.0%ni, 97.4%id, 0.1%wa, 0.0%hi, 2.2%si, 0.0%st Feb 04 21:50:44 Mem: 8169012k total, 7835824k used, 333188k free, 808400k buffers Feb 04 21:50:45 Swap: 2097148k total, 0k used, 2097148k free, 3777192k cached Feb 04 21:50:47 my server Feb 04 21:50:51 dude, excessive Feb 04 21:51:01 why I have to use CPU to to idle? Feb 04 21:51:18 14 days of uptime is pretty shit for a server :) Feb 04 21:51:22 WizardNumberNext: you don't, that's the unused cpu Feb 04 21:51:40 teotwaki: 3.6.11-rt25 hunged on kvm Feb 04 21:51:46 lemme try and find one of our other beasts, hang on Feb 04 21:52:16 kerio; no not busyloop. In Linux 2.2 it was included in idle number instead Feb 04 21:52:17 * DocScrutinizer05 yawns and heads out Feb 04 21:52:19 P.S. that's on 800MHz in 99,999% of time Feb 04 21:53:11 I have to load core to 95% to jump to next higher frequency Feb 04 21:53:23 just 4 logical CPUs ? :) Feb 04 21:53:27 please no more pastes Feb 04 21:53:27 pfft :) Feb 04 21:54:01 our lab has (multiple ?) 48 log-cpu compute servers Feb 04 21:54:16 and 65% will put it to next lower frequency (tweaked C'n'Q and cpufreq-conservative) Feb 04 21:54:48 M4rtinK: that is my home-server! I do not need anything more powerfull Feb 04 21:54:53 any top that has line-per-cpu is pretty much unusable on them :) Feb 04 21:55:14 it is overkill already - my first one was AMD Athlon 650 Feb 04 21:55:22 pretty nice, generates global Monav routing data in 5 hours :) Feb 04 21:55:27 M4rtinK: but what if you want to know precisely what the cpu #33 is doing? Feb 04 21:55:47 grep ? :) Feb 04 21:55:52 hehe Feb 04 21:56:36 kerio; WA is like "cpu idle, but there are processes waiting to run that can't run because system is waiting on data" Feb 04 21:56:49 i see Feb 04 21:57:09 For example a cp of a big file will make WA go to almost 99% Feb 04 21:57:37 or NFS on one of my 372GiB HDD Feb 04 21:57:42 But if you simultaneously run a cpu hog, WA will be 0% because the cpu will just run the cpu hog while I/O is in progress Feb 04 21:57:47 strangly SAMBA have no problem Feb 04 21:58:22 problem? Feb 04 21:58:39 DocScrutinizer05: pong Feb 04 21:59:42 strangly one of 400gb SATA HDDs is very slow with NFS (unusble for anything above few KiB), but on SAMBA there is nothing like that happening - happily swolows everything what is thrown at it Feb 04 22:01:03 other HDDs behave normaly and are faster with NFS, this one is unuable with NFS and very fast with SAMBA (but still slower, then slowest on NFS) Feb 04 22:02:42 Pali ping Feb 04 22:02:53 WizardNumberNext, pong Feb 04 22:03:20 Pali: could you enlighten me and give cmdline for 3.8rc3? Feb 04 22:03:36 cmdline? Feb 04 22:03:38 for what? Feb 04 22:03:40 I would like to see it, but cannot find cmdline for it in backlogs Feb 04 22:03:48 kernel - 3.8rc3 Feb 04 22:03:55 it is one of yours Feb 04 22:03:58 and which cmdline? Feb 04 22:04:01 for SD card? Feb 04 22:04:05 for ubi? Feb 04 22:04:12 M4rtinK: the problem with those massive multiproc systems is that you need to run something on it that can actually utilise it Feb 04 22:04:12 for rescue - from same place Feb 04 22:04:26 initrd from your server Feb 04 22:04:28 anyone which uses maemo's xchat that know if you can avoid the horrible white background under channels and nicks (assuming it doesn't fit the screen) without using the system font? Feb 04 22:04:37 M4rtinK: as far as I can tell, most schedulers start choking when you have more than 20 concurrent threads. Feb 04 22:04:38 same as for rescue os Feb 04 22:04:45 ~rescueos Feb 04 22:04:46 i heard rescue-os is http://206.253.166.96/N900/rescueOS/ Feb 04 22:04:46 I use Maemo's Quassel. Feb 04 22:04:48 teotwaki: fold *all* the homes _ò/ Feb 04 22:04:49 like.. making it black instea Feb 04 22:04:50 d Feb 04 22:04:53 ty Feb 04 22:04:58 It's not scriptable, but a good client. Feb 04 22:04:58 teotwaki: yeah, it was quite a challenge Feb 04 22:05:06 ^^ rootdelay root=/dev/ram0 Feb 04 22:05:21 The other issue is uptime Feb 04 22:06:01 23:37:53 up 500 days, 9:16, 1 user, load average: 1.00, 1.10, 1.07 Feb 04 22:06:03 as the Monav preprocessor runs only part of the computation in threads, the single-threaded overhead would make it slower Feb 04 22:06:06 :) Feb 04 22:06:11 Pali Feb 04 22:06:23 Have you seen any N900s for sale around here? Feb 04 22:06:27 Pali are all needed modules for maemo in initrd? Feb 04 22:06:39 Pali: I just wanted to highlight your nick, as everyone else is doing it Feb 04 22:06:46 Pali: hi! Feb 04 22:06:48 haha Feb 04 22:06:51 solution: cascading download->compute->compress->publish built with Python multiprocessing :) Feb 04 22:07:04 M4rtinK: have you had a look at 0mq? Feb 04 22:07:09 WizardNumberNext, in which initrd? Feb 04 22:07:30 M4rtinK: logic would dictate that you first get something to do with 48 cpus, and then you get the machine to run it Feb 04 22:07:31 Aoyagi, where? Feb 04 22:07:44 kerio: that's not how universities work. Feb 04 22:07:57 for anyone interested: https://github.com/M4rtinK/modrana-data-repository :) Feb 04 22:08:01 Pali: this one: https://atrey.karlin.mff.cuni.cz/~pali/linux-n900/rescueOS-initrd-3.8-rc3.img Feb 04 22:08:07 that is old Feb 04 22:08:19 modules in that initrd are old Feb 04 22:08:19 Pali: In the reeking town you happen to study in, according to that site. Feb 04 22:08:35 WizardNumberNext, compile kernel yourself Feb 04 22:08:41 so where are new ones and matching kernel? Feb 04 22:08:42 and put them into initrd Feb 04 22:08:51 kerio: I asked the lab guys if they have some hardware I can use to host a Monav data repository for modRana Feb 04 22:09:13 and they were so kind to give me access to one of the compute servers Feb 04 22:09:21 I would do that, when I would compile my cross-compiler Feb 04 22:09:32 Aoyagi, I do not know about any shop in SK/CZ which has n900 Feb 04 22:09:50 Pali: Yeah, that's what I found out too. Feb 04 22:10:12 M4rtinK: nice, and impressive. Feb 04 22:10:29 Except some used stuff, that has displays in poor condition. Or some other intolerable faults... Feb 04 22:10:40 teotwaki: thanks ! :) btw, mq0 also looks interesting Feb 04 22:10:56 Aoyagi, you can watch this site http://mobilni-telefony.heureka.cz/nokia-n900/ Feb 04 22:10:58 probably a bit overkill for what I'm doing though :) Feb 04 22:11:41 an improved Monav preprocessor would probably help more Feb 04 22:11:51 M4rtinK: Most of our 100 servers or so are either quad or octo cores, half of that runs xens with on average 30 VMs each. The other half of the servers run native distribs with RT kernels and handle roughly between 4 and 12 always-on processes. Feb 04 22:12:04 Pali: I did, then I found out mp.cz has it, but it was some kind of "mistake" on their part. Feb 04 22:12:31 teotwaki: impressive Feb 04 22:12:33 M4rtinK, do you know if it is possible to buy new n900? Feb 04 22:12:45 somewhere? Feb 04 22:12:49 new ones? Feb 04 22:13:04 yes new Feb 04 22:13:07 Not necessarily new one. :D Feb 04 22:13:13 teotwaki: I'm not really sure how many machines we have, but the one I'm using is the 4th one Feb 04 22:13:52 M4rtinK: to be honest, I'm just repeating numbers. I build the solutions the runs on that hardware, I have no clue :P Feb 04 22:14:11 our lab does Natural Language Processing (http://nlp.fi.muni.cz/en/nlpc), so probably many :) Feb 04 22:14:22 Pali: I'll check Feb 04 22:15:05 M4rtinK: that's pretty sweet Feb 04 22:15:36 Pali: check this: http://www.zbozi.cz/vyrobek/nokia-n900-black/ Feb 04 22:15:44 two even report "in stock" Feb 04 22:16:03 yeah, cuz that really looks like an N900 Feb 04 22:16:04 I bet those are Lumias Feb 04 22:16:21 bah, can't even check my assumptions re repo becaues it limits my dl to err nothing, yay Feb 04 22:16:39 Pali: I almost died... Feb 04 22:16:41 Ah, it's actually N9 Feb 04 22:17:01 yes Feb 04 22:17:03 N9-00 Feb 04 22:17:25 I tried it and after few minutes N900 simply died and didn't want to turn on. Feb 04 22:17:46 strangly battery was reason Feb 04 22:17:54 Pali: also http://www.mobilyodalese.cz/nokia-n900/d-69905/ Feb 04 22:17:58 WizardNumberNext, that is old version without battery charging Feb 04 22:18:02 in kernel Feb 04 22:18:16 Aoyagi, see ^^^ Feb 04 22:18:33 "na dotaz" :/ Feb 04 22:18:39 I know that, but after rawly 3 hours of being on USB it should be (almost?) fully charged Feb 04 22:18:48 As far as I know, it means "ask and we'll tell you it will never be here" :D Feb 04 22:19:16 I'll just have to find a good looking used one. Feb 04 22:19:27 rawly? Feb 04 22:19:44 Aoyagi, you can buy from amazon Feb 04 22:19:51 Pali: http://mobily.aukro.cz/listing.php/adSearch?string=n900&category=107156&minResult=1&utm_source=seznam&utm_medium=sklik Feb 04 22:20:03 Aoyagi: I do not actually trust that, but some seller from HK on eBay have some "brand new" Feb 04 22:20:25 Aoyagi: yeah, you never know in those small shops Feb 04 22:20:39 merlin1991: We hopefully get more bandwidth tomorrow. Feb 04 22:20:45 Well, I guess asking in the UK couldn't hurt. Feb 04 22:20:45 they might not be even operating anymore Feb 04 22:21:08 other times they work just fine Feb 04 22:21:12 never have seen any new in UK Feb 04 22:21:18 especially in London Feb 04 22:21:42 warfare: you could also verify that for me on the server, I'd need md5 hashes of all files withint dist/fremantle-1.3/free from extras and extras-devel Feb 04 22:21:56 the other way is to dl them and make hashes locally .) Feb 04 22:22:15 ABSOLUTE RECORD: 3125mV and it is alife! Feb 04 22:22:42 never have seen it so low! Feb 04 22:24:19 merlin1991: http://privatepaste.com/b65516c41a Feb 04 22:28:12 so older packages seem to be unaffected by the hash mismatch error: Feb 04 22:28:15 http://privatepaste.com/883d83e9a4 Feb 04 22:28:39 BTW: Fetched 5224B in 25min19s (3B/s) ??? :D Feb 04 22:29:31 warfare: it's funky, a few lines in the Release file have the right checksums Feb 04 22:29:34 but most are wrong Feb 04 22:31:10 hm wait Feb 04 22:34:46 WizardNumberNext: you can get as low as 2800, but shit starts malfunctioning at that point Feb 04 22:35:30 I never got below 3200mV and suddenly 3125mV Feb 04 22:35:53 warfare: the md5sums on the indexes seem to be right Feb 04 22:36:14 (at least on the relevant arm ones) Feb 04 22:36:38 now I wonder why we get hashsum missmatches with apt Feb 04 22:41:30 warfare: some reindex of the repo / import / whatever must have run, because the Release files fits with the package indexes, but it does not on my mirror, so somewhere between the 26th Dez and now it got fixed again Feb 04 22:45:17 merlin1991: do Packages and Packages.gz match? Feb 04 22:45:30 kerio: I didn't download the files because they interupt Feb 04 22:45:51 I only said that the hashes on those now fit with the Release file (they didn't some time ago) Feb 04 22:46:01 i see Feb 04 22:46:36 well, updating works, it's some packages that report a hash mismatch Feb 04 22:53:34 well with my mirror even updating fails Feb 04 23:12:36 * WizardNumberNext should call himself SharkNumberNext Feb 04 23:21:07 what do you think about communi? is irggu better or worse? Why? Feb 04 23:22:46 merlin1991: sorry, I'm completely lost on what you're suggesting resp planning to do about that hashsum stuff Feb 04 23:25:23 DocScrutinizer05: the hashsums on the index files seem to be okay, the only thing left is to check the actual Package indexes against the .debs in the pool directory Feb 04 23:25:58 [2013-02-04 22:17:26] n900-dk: are you trying to use rmo? Feb 04 23:26:00 [2013-02-04 22:18:17] yeah, it downloads but gives 'Hash Sum mismatch' Feb 04 23:26:14 [2013-02-04 22:19:24] n900-dk: I'm also getting hash mismatch Feb 04 23:27:02 really seems to affect only recent packages Feb 04 23:27:05 I never got so far as mismatched sum - never actually been able to pull Packages.gz Feb 04 23:27:44 DocScrutinizer05: you scared me using my N900 Feb 04 23:28:00 heh? Feb 04 23:28:47 I've put WizardNumberNext (vibra again) on N900 on communi Feb 04 23:29:35 I am growing 2nd set of pernament teeths, so I should be called SharkNumberNext, then Feb 04 23:29:45 I hope that's not meant to make sense to me Feb 04 23:31:46 it doesn't make any sense to me either, but I just pulled one of my pernament teeth and there's another pernament replecement just behind it - looks like my jaw want to become jaw of shark Feb 04 23:32:12 merlin1991: i'd be more interested in the hashsum-error_for_dummies 101, about which server has and which hasn't errors, and maybe then we can think about the why Feb 04 23:32:46 DocScrutinizer05: I'm still working on the where bit Feb 04 23:33:07 hence my remark about comparing Package indexes to .deb files Feb 04 23:33:30 yeah fine Feb 04 23:33:48 still doesn't tell me which of our 3 mirrors has errors now and which hasn't Feb 04 23:34:10 okay 101 sttyle Feb 04 23:35:06 r.m.o has clean distributions files, but I don't know if the hash sums for the .debs are right, my mirror even has foobar distribution files, and no idea about skeiron Feb 04 23:35:26 aha Feb 04 23:35:44 obviously rmo has errors as well Feb 04 23:35:50 also the distribution files on r.m.o are newer than on my mirror Feb 04 23:36:19 i think i got some errors from skeiron few days ago Feb 04 23:38:45 so nobody ever systematically checked which repo/mirror has which problems? Feb 04 23:38:59 since 4 weeks? Feb 04 23:39:41 well since the base was corrupted all mirrors have some problems Feb 04 23:39:50 e.g. by trying to install an old and a more recent package from all 4 possible sources Feb 04 23:39:56 r.m.o is the canonical sync point and it has to be fixed first Feb 04 23:40:46 actually, it should all be fixed Feb 04 23:40:47 merlin1991: we got the mirrors for backup reasons, so when r.m.o breaks or has other problems, we can resort to a working mirror Feb 04 23:41:09 if your backups aren't backups, there's no point in using them Feb 04 23:41:21 just shut down the whole infrastructure, figure it out, set it back up. Feb 04 23:41:23 DocScrutinizer05: if the canonical sync point goes bad before you notice your backup goes bad Feb 04 23:41:36 and all our backups are in a semi bad state Feb 04 23:41:47 so we have to fix r.m.o and then resync to a sane state Feb 04 23:41:50 says who? Feb 04 23:42:07 people doing backups Feb 04 23:42:17 so people are backuping the backups of the backup Feb 04 23:42:17 DocScrutinizer05: says me Feb 04 23:42:22 [2013-02-05 00:35:06] r.m.o has clean distributions files, but I don't know if the hash sums for the .debs are right, my mirror even has foobar distribution files, and no idea about skeiron Feb 04 23:42:28 Did you guys read Dr Seuss perchance? Feb 04 23:42:42 DocScrutinizer05: and what exactly does that say? Feb 04 23:42:48 * DocScrutinizer05 heads out Feb 04 23:42:59 also the distribution files on r.m.o are newer than on my mirror Feb 04 23:43:15 The backupers backup the backup. But the backup backupers were also backed up by the backed up backup backupers. Feb 04 23:43:18 r.m.o cchanged to the better since dez 29 which is the last valid sync on my server Feb 04 23:43:59 but since apt-mirror only runs wget without any postup checks my mirror which has the state from 29th dez is full on corrupted Feb 04 23:44:37 and since there is no versioning or anything I can't roll back to a sane point Feb 04 23:44:49 hence my fix r.m.o resync Feb 04 23:45:01 DocScrutinizer05: ^ Feb 04 23:45:02 because of missing incremental-rsync Feb 04 23:46:20 chem|st: well that would have been helpfull :) Feb 04 23:46:52 DocScrutinizer05: how do I update a factoid? Feb 04 23:47:08 gotta remove my server from the list because it is full on foobar Feb 04 23:47:27 kerio: ping Feb 05 00:00:21 well, looks like we need something more, then simple apt-mirror Feb 05 00:01:04 maybe something like my backup script, but we would un-backup script for it as well, which doesn't exists yet Feb 05 00:03:19 actually using my backup script, you can still use apt-mirror, but in case of problems it tedious date-time removing from filename (in backup) Feb 05 00:05:13 in any case r.m.o needs a sanity check Feb 05 00:05:29 but at least the Release file seems to be in order (again) Feb 05 00:06:37 http://paste.debian.net/231819/ Feb 05 00:06:58 you can have a look on my simple backup script Feb 05 00:07:15 it is most reasonable version up to date Feb 05 00:07:58 oww, i now just found out that this n900 has no metal holder in microsd socket Feb 05 00:09:00 ketas: not too nice Feb 05 00:09:36 doesn't seem like anything is broken... Feb 05 00:10:01 how to get one off anyway Feb 05 00:10:26 i wonder where to find replacement Feb 05 00:11:50 maybe eBay but, I didn't see any till something like 25GBP Feb 05 00:12:13 well, unless you want to tape card there Feb 05 00:12:18 sounds shitty solution Feb 05 00:12:28 very damn shitty Feb 05 00:12:37 intermittent connection fun Feb 05 00:13:37 but actually not uch bad Feb 05 00:14:23 using some sponge-like substance and tape would solve problem Feb 05 00:14:47 this sd holder must be sold in some catalogue i hope Feb 05 00:14:56 small chunk of foam below contact and wrap everything with tape - voila - should work Feb 05 00:15:07 maybe nokia have it Feb 05 00:15:22 maybe nokia service has busted pcb? Feb 05 00:15:28 let me take a look on service manuals - maybe I would find it Feb 05 00:15:39 bit old phone, though Feb 05 00:15:54 they have PCBs as well - I am quite sure about that Feb 05 00:16:59 however i'm sure such holder is actually used in other pcb's somewhere Feb 05 00:17:09 where one could try getting slider off Feb 05 00:17:17 how does one get it off anyway Feb 05 00:17:55 X5210 in Nokia Feb 05 00:18:12 at least looks like Feb 05 00:19:12 I got it of on SIM and SD accidentially, but it is easy to do it both ways Feb 05 00:19:24 putting it back isn't hard either Feb 05 00:19:39 i wonder if previous owner still has it Feb 05 00:19:45 merlin1991, please test clock-ui on your n900 (we need to test on thumb and non thumb versions too) Feb 05 00:19:50 it was year ago... Feb 05 00:19:53 yes, in rubbish omewhere Feb 05 00:20:00 there is very very ugly hack for time cp dialog Feb 05 00:20:18 and why he took it off and never put it back? Feb 05 00:20:35 becase that cp plugin is closed and execute function needs some parent gtkwidget Feb 05 00:20:52 maybe it was accident and it got wings on its way to ground Feb 05 00:21:01 Pali: should I compile it, or do you have a specific binary in mind? Feb 05 00:21:15 and because it is qt application we do not have gtk top level widget Feb 05 00:21:30 merlin1991, better compile Feb 05 00:21:39 we need to be sure that it is working fine Feb 05 00:21:41 k, will do that tomorrow then, gonna head to bed soon Feb 05 00:21:45 I have binary from madde Feb 05 00:21:59 * WizardNumberNext wonders why Nokia have mixed gtk and qt in OS with not so much RAM Feb 05 00:22:01 and this non thumb binary working fine on cssu-thumb Feb 05 00:22:39 WizardNumberNext, nokia did not mixed it Feb 05 00:22:45 nokia used only gtk for all applications Feb 05 00:22:57 3rd developers used qt Feb 05 00:23:00 so why other people did it? Feb 05 00:23:13 do not ask me why Feb 05 00:23:17 while gtk is good enough Feb 05 00:23:25 it does the job for me Feb 05 00:23:45 WizardNumberNext, I think that because 3rd devs developed also for symbian or harmattan Feb 05 00:23:46 actually not on PC - never liked anything GTK based there Feb 05 00:23:58 and Qt is on all 3 platforms Feb 05 00:24:00 you might be right Feb 05 00:24:28 but why is that qt applications are much harder to run than gtk Feb 05 00:24:32 and you have button for "compile for XYZ" in QT creator Feb 05 00:24:47 never tried to write anything in gtk or qt Feb 05 00:24:48 because QT eats more resources? Feb 05 00:24:52 ketas, which app? that clock-ui? Feb 05 00:25:22 nah, in general, and in pc i meant Feb 05 00:26:09 because qt is c++, gtk is c Feb 05 00:26:17 GTK is believed to be ligher, but I am not in subject recently, so it could have changed in those several years spend on Debian+KDE Feb 05 00:26:23 why do i feel that i don't fit into this world? Feb 05 00:26:33 see specification of c++ and what is there and then see c Feb 05 00:26:42 Pali: which explains why QT is more resource-hungry as well Feb 05 00:26:53 yes, right Feb 05 00:27:30 Pali: you can extend both, but starting is much harder on plain C, then plain C++ Feb 05 00:27:53 s/starting/starting to extend/ Feb 05 00:27:53 WizardNumberNext meant: Pali: you can extend both, but starting to extend is much harder on plain C, then plain C++ Feb 05 00:27:58 each qt class has table for virtual functions Feb 05 00:28:11 you have no control for that Feb 05 00:28:39 at least it runs and looks modern-enough Feb 05 00:28:50 "I saw C++ coming and read the book - or at least tried to read the book - there's a dent in the wall behind my piano, where the book hit the wall - Improvements to C should make things easier not more complicated, I thought." Feb 05 00:28:52 in gtk (c) you need to write table yourself (if using inherits functions) Feb 05 00:29:05 GNOME/GTK in my times looked realy crappy Feb 05 00:29:08 but more my apps/plugins in maemo did not use that Feb 05 00:30:20 Pali: I really don't know what is better: C or C++ - it might be matter of preference Feb 05 00:30:28 never wrote anything in C++ Feb 05 00:30:37 ( from http://erlang.org/pipermail/erlang-questions/2013-January/071949.html ) Feb 05 00:30:51 always C, but to simple progs to be proud and priase myself Feb 05 00:31:06 never tried erlang Feb 05 00:31:15 http://www.leancrew.com/all-this/2011/12/more-shell-less-egg/ was also related funny reading Feb 05 00:37:06 WizardNumberNext: easy -> Python :) Feb 05 00:37:35 but yeah, probably not for the GUI libraries themselves though :) Feb 05 00:38:03 Python should be my favourite as it similar to C and can be both ompiled and run like script Feb 05 00:38:05 python is too weird for one with c-like background Feb 05 00:38:30 I can swollow anything what is reasonable and have some good base Feb 05 00:38:40 I think about learning ADA for fun Feb 05 00:38:47 WizardNumberNext: well, IIRC it is not that easy to just compile it :) Feb 05 00:38:50 better perl Feb 05 00:38:54 but this one is quite good lanuage Feb 05 00:39:03 like the whole program Feb 05 00:39:11 * ketas pats Pali Feb 05 00:39:20 perl for phones Feb 05 00:39:53 nothing is just easy - you are capable of doing it or not, then you can learn it or never try Feb 05 00:39:55 you have both gtk and qt bindings in perl Feb 05 00:40:14 but you you can use a limited syntax for performance critical stuff (was it Cython ?) and compile that Feb 05 00:40:20 trying to learn isn't enough - my friend is doing just that for last 3 years - no efects Feb 05 00:40:27 or just runn the Python code ass a fallback Feb 05 00:40:51 I asked him to write program, which would find 1Mi prime numbers - he never got back to me with it Feb 05 00:41:16 and writing C extensions should be also quite easy (that's basically what the Python stdlib is based on) Feb 05 00:42:05 bindings are an important aspect when choosing a programming language Feb 05 00:42:33 when you wan't to write a graphical application, you choices are actually quite limited... Feb 05 00:42:53 I never was interested in graphical apps Feb 05 00:43:09 what I do is simply make my life more efficient Feb 05 00:43:20 then your choice is much less restricted :) Feb 05 00:43:51 so do I Feb 05 00:44:06 instead of spending years on simple task, I spend 4 hours on complicated task and 40 minutes on even more complicated task and have results just in few minutes Feb 05 00:44:36 but some things just kinda require a GUI to use (navigation system, manga/comic book reader, rephotography app)... Feb 05 00:45:14 an ascii based ones would be also kinda cool though :) Feb 05 00:45:54 like it was with my first prime finding prog - 4h for writinig it, next hour for finding it is extremaly slow, 40m for optimalization and slightly over 10m for 1Mi prime numbers Feb 05 00:46:56 it is shame it is lost with my first HDD Feb 05 00:47:08 oh, it was on 40GB, so second Feb 05 00:47:50 maybe I would be able to recover it from it, but it might impossible after running proxy on same HDD Feb 05 00:49:35 * WizardNumberNext needs to move to Xen and 149GiB HDD on server Feb 05 00:52:43 time to go :) Feb 05 00:52:45 bye :) Feb 05 00:57:38 see yea M4rtinK Feb 05 01:17:29 i am getting a hash sum mismatch. what should i do? Feb 05 01:21:06 rage Feb 05 01:21:08 also Feb 05 01:21:32 please report on which package you try to install and which repos you have enabled Feb 05 01:21:52 report where? Feb 05 02:02:57 Aoyagi_joytop, you want to buy n900? Feb 05 02:03:08 I have 3 for sale atm, with varying condition Feb 05 02:03:21 tell me your budget and I'll tell you what you can buy for that Feb 05 02:03:28 all have reinforced usb port, mind you Feb 05 02:04:12 as for repos - so, tmo reports are not only working slow, but they're also giving files with wrong hash, due to some fckup? Feb 05 02:04:32 which means that everyone downloading smth will start another download, as soon as he finished last one... Feb 05 02:04:56 thus, load on rmo never ends, as everyone is downloading faulty files and redownloading them again, and again, and... Feb 05 02:05:00 nice Feb 05 02:05:00 Estel_: the stable repo doesnt have the hash problem afaict Feb 05 02:05:11 who uses stable repo? Feb 05 02:05:18 me Feb 05 02:05:49 even completely non-power, not to say "dumb" users have devel enabled, most of the time - by dumb I mean ones that doesn't know a shit about maemo internals, never been on irc or tmo or... Feb 05 02:05:59 well, you're minority ;) Feb 05 02:06:31 the problem is that load from "dumb" users will never end, as their devices will try to redownload faulty files infinitely Feb 05 02:06:42 damn Feb 05 02:06:44 that sucks Feb 05 02:06:56 maybe the repo will get hammered Feb 05 02:07:21 I'm quite surprised no one thought already, that those fckd files with wrong hash'es are having their share in repos being hammered to death Feb 05 02:07:31 anyway **** ENDING LOGGING AT Tue Feb 05 02:59:59 2013