**** BEGIN LOGGING AT Mon Sep 23 02:59:59 2013 Sep 23 11:16:38 moo Sep 23 11:31:31 µ Sep 23 11:41:13 ~dict µ Sep 23 11:41:15 could not find definition for µ Sep 23 11:41:22 ~uselss Sep 23 11:41:30 ~useless Sep 23 11:41:31 * infobot starts crying and hides from docscrutinizer05 in the darkest corner of the room. :( Sep 23 14:04:51 so anyone knows if the advanced power program adds any benefit to power management? Sep 23 14:07:55 is there a way to import scratchbox deb signatues for their apt repo? Sep 23 14:16:36 oooaaaaaauuu starts to annoy me Sep 23 14:17:33 honestly, waiting almost 10 min and nobody answers! unbearable on IRC! :-S Sep 23 14:18:23 Doesn't that remind you of a similar case that applied to someone else? :) Sep 23 14:18:38 err, there been several Sep 23 14:18:49 nobody in particular comes to mind Sep 23 14:19:54 * psycho_oreos was thinking of the most recent one. Sep 23 14:20:10 but I seem to recall educating oooaaaooo some two days ago that he shouldn't ask questions when he's not prepared to wait for somebody answering a few hours later maybe Sep 23 14:21:18 *shrugs* I didn't recall that (or maybe I've completely missed it). I guess maybe a /topic change telling people should be expected to wait. :) Sep 23 14:21:32 psycho_oreos: sorry, such stuff gets stored to my archive with very few cross links Sep 23 14:23:01 so unless it shows up by the same nick hitting my radar with same behaviour, odds are it gets lost in archive forever Sep 23 14:23:32 DocScrutinizer05, I guess hence its probably easily for you to get irritated. :) Then again I agree that people (especially inquirers) are expected to wait, otherwise pay for priority support. Sep 23 14:23:48 I'm dealing with ~2k nicks on IRC, in those channels I moderate only Sep 23 14:24:31 I pondered a nick-annotation function extension for IRC clients already Sep 23 14:25:33 right click on a post, from context menu select "add this quote to user's rating" Sep 23 14:25:35 haha, that would sort of be going a bit too far. :) At any rate I would personally vote for /topic change to tell inquirers to wait. If they couldn't wait, that's fine they won't get their answers. Sep 23 14:26:36 psycho_oreos: adding too much general rules to topic doesn't serve a purpose Sep 23 14:27:31 particularly since /topic length has a technical maximum length Sep 23 14:27:42 DocScrutinizer05, *shrugs* true if the inquirers usually ignore /topic. Then again, it may help. Sep 23 14:27:46 eeek tautology Sep 23 14:28:02 o.O yeah there's that as well. (maximum length for /topic). Sep 23 14:28:26 Maybe onjoin chanserv greet telling them to be expected to wait. Sep 23 14:29:27 robotanarchy, also considered about asking the same question in #scratchbox ? Sep 23 14:29:48 I could set up an autoresponder on regex ".*\?" --> "WAIT til somebody answers! we're not on 24/7 standby to do your 1st level support in less than 5 min" Sep 23 14:30:18 psycho_oreos: ah there's a channel for that? thanks xD Sep 23 14:30:32 it's amazing how much the IRC of maemo etc. is populated in 2013 Sep 23 14:30:38 That would also work. :D Sep 23 14:31:22 robotanarchy, yeah I recall there was. Anyway I checked via /msg alis list *scratchbox* and there was one. Seems like only 10 people/bots are in it. Sep 23 14:32:55 probably questions regarding SB rather go to some ML, since on IRC the SB support is scarce Sep 23 14:33:35 the SB used in maemo is not even in maintenance mode anymore since at least 3 years Sep 23 14:35:16 afaik they orphaned SB2 now Sep 23 14:35:27 we're using SB(1) Sep 23 14:35:45 DocScrutinizer05: ok thanks, if I have a really big important question, I'll put it on the ML :) all the tutorials are for sb1 and I haven't done this before, so I'll stick to that, too Sep 23 14:36:34 I've also thought about sort-of building my own crosscompiling toolchain, but then again I only want to compile a few packages^^ Sep 23 14:37:06 SB is a PITA but once you got it set up it kinda works Sep 23 14:37:33 ~usb Sep 23 14:37:34 hmm... usb is Unusable Serial Bus. Useless Serial Bus. Ugly Stupid Bus Sep 23 14:37:45 and it seems every possible problem arising from SB in maemo development got sorted meanwhile Sep 23 14:37:46 ~usbfix Sep 23 14:37:46 well, usbfix is http://talk.maemo.org/showthread.php?t=75920 - and **NEVER** use epoxy (unless you want to seal your device for underwater) Sep 23 14:38:33 I never heard about "importing SB sig to apt" Sep 23 14:39:14 robotanarchy, if you have issues setting up scratchbox. I was just thinking it maybe an easier solution (if possible) to use one of those pre-made VM with scratchbox already setup within it. Sep 23 14:40:00 I'm guessing it sounds like one wants to install SB through apt. Then there's something like apt-key for instance. Sep 23 14:40:03 DocScrutinizer05: I guess their packages aren't signed. would be good practice though, since we build all the packages with that software (openssl, ssh, ... for starters) Sep 23 14:40:32 nah, robotanarchy's question alludes to some signature of .deb packages SB allegedly does and that apt is not aware of or can't check. I never heard of SB doing any such signatures Sep 23 14:40:44 psycho_oreos: yes, I do it with apt, have already set up a i386 debian wheezy LXC VM for that and I guess it will work out, just wondered whether it was possible to verify the packages Sep 23 14:42:03 robotanarchy, well I guess DocScrutinizer05's answer seems to be more precise there then. Sep 23 14:42:23 thanks anyway :) Sep 23 14:43:24 robotanarchy: you have full control over the complete lifecycle of your homemade .deb. signing them won't help for anything Sep 23 14:43:43 i meant, if the scratchbox debs were signed Sep 23 14:43:50 they are not Sep 23 14:43:52 it would help for man in the middle attacks etc Sep 23 14:44:01 yes that's clear now Sep 23 14:44:15 I would notice any man in the middle of my USB cable ;-D Sep 23 14:45:09 Signing packages (especially deb) reminds me too much about 'harmattan'. :p Sep 23 14:45:13 lol sure, I mean: when I set up my VM for debian package building, first I need the scratchbox binaries (from the scratchbox website, in deb form per apt-get). it would be nice if these debs from scratchbox.org were signed Sep 23 14:45:31 after all you're not supposed to publish your homemade .deb to the general public - and even when you do, those who are receiving them don't have any valid pubkey to verify your private signature Sep 23 14:45:54 ooh, those Sep 23 14:46:06 those are "secured" by md5sum Sep 23 14:46:13 afaik Sep 23 14:47:13 DocScrutinizer05: Yes, debian packages have integrated md5 sums. I guess you get my point :) Sep 23 14:48:02 If say in theory scratchbox were to be that popular, I'm sure the devs would have considered signing deb packages. Sep 23 14:48:03 I just can't do anything about that, since we don't know about the original sb binaries either, since we didn't build them Sep 23 14:48:39 we're providing them "as is" via https:// if you like Sep 23 14:49:05 that's as good a MITM attack prevention as it gets in this case Sep 23 14:49:30 DocScrutinizer05: that would be better, yes :) Sep 23 14:50:22 ~whois scratchbox.org Sep 23 14:50:23 whois: user 'scratchbox.org' does not exist. Sep 23 14:50:31 lol Sep 23 14:52:24 we inherited hosting of scratchbox.org from original maintainers, adn we provide the content "as is", we can't do any further signing on those "blobs", except https transfer. I dunno if scratchbox VM has a proper SSL cert for https, might check eventually Sep 23 14:53:49 MEH, no https on sb.org Sep 23 14:54:00 thedead1440: ^^^^ could you check that please? Sep 23 14:54:40 DocScrutinizer05: you could also provide sha512 sums on the blobs on another site, so users could check them (I might update/write a wiki article about scratchbox setup when I have mine, so I'd put that there) Sep 23 14:54:44 thedead1440: I'll eventually try to get a semi-decent SSL cert for it Sep 23 14:55:52 robotanarchy: please ask thedead1440, he's our "new" sb vm maintainer Sep 23 14:56:11 DocScrutinizer05: ok :) Sep 23 14:56:13 of course he can create sha512 sums and provide them on site Sep 23 14:57:19 thedead1440: until proper cert from CA, please ask warfare to provide self-signed cert for sb vm Sep 23 14:58:27 please send a mail to techstaff, with the requirement/change-request Sep 23 14:59:02 I won't keep track of this issue Sep 23 15:11:38 Is there some known bugs in N900 media player's mp3 tag support? I have some MP3s on the N900 and some of them are listed as unknown artist, though I can see there are id3v2 tags Sep 23 15:19:11 dunno Sep 23 15:19:18 try musorgsky Sep 23 15:19:34 id3 editor Sep 23 15:20:11 if there's a bug, then it's in trackerd Sep 23 15:20:31 not in media player Sep 23 15:20:56 ~tracker Sep 23 15:21:03 grr Sep 23 15:21:14 ~listkeys tracker Sep 23 15:21:16 Factoid search of 'tracker' by key (15): ftp mirror tracker ;; soundtracker ;; sarah tracker ;; jdtracker ;; timetracker ;; voodoo tracker ;; request tracker ;; jsetitracker ;; asterisk issue tracker ;; bugtracker ;; stracker ;; trackerreset ;; keg tracker ;; impulse tracker 3 ;; laptop tracker. Sep 23 15:21:45 ~trackerreset Sep 23 15:21:46 extra, extra, read all about it, trackerreset is "tracker-processes --hard-reset" in Terminal Sep 23 15:22:15 ~factinfo trackerreset Sep 23 15:22:15 trackerreset -- created by MohammadAG <~MohammadA@62.219.120.20> at Wed Jun 16 21:53:01 2010 (1194 days); it has been requested 8 times, last by DocScrutinizer05, 29s ago. Sep 23 15:30:30 DocScrutinizer05, I've edited the tags with "id3v2" Sep 23 15:31:04 commandline tool, that's why I like it. Sep 23 15:31:11 sorry, hardly anything *I* can do about it Sep 23 15:32:47 I just can tell you that mediaplayer does NOT read id3 tags, it's tracker that does, when scanning your media files. this scan happening only now and then, triggered by a set of conditions Sep 23 15:33:15 DocScrutinizer05, That's news to me Sep 23 15:34:07 I assume one of the triggering conditions is exiting from USB mass storage mode... Sep 23 15:34:18 err, nope Sep 23 15:34:45 changes in files on any of the media storage volumes are, as seen by inotify Sep 23 15:35:10 and there are options to disallow any indexing e.g. when on battery Sep 23 15:35:53 trackerreset usually triggers a full indexing run Sep 23 15:36:13 which can take like 30 minutes, on huge mp3 volumes Sep 23 15:36:38 I have some ~25-30files Sep 23 15:36:49 should take only seconds then Sep 23 15:37:25 So: tracker-processes --hard-reset will force re-indexing if I understood correctly Sep 23 15:38:14 :nod: Sep 23 15:38:57 touching any file should usually also force reindexing of prolly that file only Sep 23 15:39:26 maybe only as long as trackerd "sees it" via inotify Sep 23 15:40:12 there are watched dirs and indexed dirs, in tracker.cfg Sep 23 15:41:03 the watched ones are prolly the ones monitored by inotify, while the indexed ones are those who are only scanned when a "huge scan" gets triggered by e.g. reboot or whatever Sep 23 15:41:29 tracker is a pity, and buggy too Sep 23 15:41:35 pta even Sep 23 15:41:44 damn! PITA Sep 23 15:43:12 On a positive side: THe player seems to support the embedded images with mp3 (or its the tracker again), even supports png. Either it masks transparency as black or it supports it, but it's better than sonos. With sonos transparent equals to white -> ugly Sep 23 15:44:30 might be thumbnailerd Sep 23 15:44:51 creating .coveralbumart Sep 23 15:45:06 or whatever the name of those files Sep 23 15:45:10 iirc Sep 23 15:46:44 meh, scrap that Sep 23 15:47:07 trackerd seems to store the coveralbumart inside the db Sep 23 15:49:23 umm no, /home/user/MyDocs/.sounds/.mediaartlocal Sep 23 15:50:51 and /media/mmc1/allofmp3/C/Cure, The/Seventeen Seconds/.mediaartlocal Sep 23 18:14:22 I wonder if there's a way to "wake" camera-ui and wait for the resulting picture (if any) via D-Bus or API or whatever Sep 23 18:14:47 I can't find a way from an app to call the camera app and if it takes a picture get the the filename of it Sep 23 18:41:05 afaik there's none Sep 23 18:41:38 you might be better off with nicocam aka cameraui2, from CSSU Sep 23 18:42:22 or take picture with generic means, there been several mini-howtos on that Sep 23 18:42:57 gstreamer command, building your own pipe Sep 23 18:45:18 or you actually want to use cameraui with all the features like manual trigger button etc, then you could simply inotify-watch the DCIM dir for any new file showing up Sep 23 18:47:27 starting the cameraui (rather: unhiding the window, since cameraui is preloaded and always running afaik) should be feasible with standard means afaik, similar to starting any other app that has a .desktop Sep 23 18:53:44 well what I didn't want to do is make my own camera-ui (if I use gstreamer I would have to kinda do that) Sep 23 18:54:12 there's a dbus signal to tell camera-ui to go to the top Sep 23 18:54:37 but I would have also need to detect if the user didn't want to take a picture Sep 23 18:55:26 why? Sep 23 18:55:42 to stop looking at the DCIM dir Sep 23 18:55:47 hmmm Sep 23 18:56:18 check dbus for signals flying by when closing c Sep 23 18:56:23 cameraui Sep 23 18:56:26 yeah I guess Sep 23 18:57:49 indeed this function "macro" is missing a lot in a number of places Sep 23 18:58:13 e.g. in avatar assigning in contacts Sep 23 18:58:58 WTF start camera, take picture, then search DCIM manually in contacts to assign that snapshot to the contact avatar? pfff Sep 23 18:59:49 mbarcode also created own GUI for camera Sep 23 19:00:39 quite a number of other apps as well would need or at least could use that genuine camera function Sep 23 19:01:41 I almost could bet Qtm comes with an own widget that provides camera function but doesn't use the standard cam GUI either Sep 23 19:03:46 maemo failed epically to support/provide those toolkit functions for "take picture", "make a note", "draw a sketch", etc Sep 23 19:04:14 "record an audio sample" Sep 23 19:04:58 "send an email"(?) Sep 23 19:05:07 "send an SMS" Sep 23 19:05:43 terribly poor API design Sep 23 19:08:09 Sc0rpius: probably you building a generic "shoot a picture" lib (or whatever, daemon, executable, dunno) would benefit all maemo community Sep 23 19:11:58 you will want to find out how to pass command line options to such executable, to define operation modes (macro, night mode, flash or no flash, etc pp) and the file where user finally will find the result. The executable should return OK (0) if everything worked, -1 for E_USER_ABORTED, < -1 for other errors (like "no free storage" or whatever) Sep 23 19:17:24 but like a wrapper of the camera-ui app? Sep 23 19:18:12 and I totally agree about epic fail from Maemo to provide an API for standard mobile operations. Sep 23 19:22:43 yep, wrapper to whatever hides behind standard cameraui Sep 23 19:25:23 alternatively nicocam could (or already does?) provide a more generic API for such stuff, but obviously that doesn't help for non-CSSU users - unless you install nicocam as *alternative* in parallel to the stock cameraui Sep 23 19:28:46 yeah and I don't really want that Sep 23 19:28:54 same for FCamera / FCam Sep 23 19:28:57 that also provides an API Sep 23 19:30:04 :nod: Sep 23 19:30:09 wrapper it is, then Sep 23 19:30:15 should be feasible Sep 23 19:31:28 with some dbus and inotify magic, plus prolly some gconf messing around for the modes (don't forget to reset to status before using the function, as soon as possible) Sep 23 19:33:09 you need such wrapper anyway, for whatever been the reason you initially asked. Just make it a stand-alone binary wrapper that can get used by other apps as well Sep 23 19:33:43 s/binary wrapper/wrapper executable/ Sep 23 19:33:44 DocScrutinizer05 meant: you need such wrapper anyway, for whatever been the reason you initially asked. Just make it a stand-alone wrapper executable that can get used by other apps as well Sep 23 19:36:04 Folks when packaging for maemo. you don't need to login into scratchbox? I mean you do the sb-conf select FREMANTLE_ARMEL and all the rest from outside ? Sep 23 19:36:53 I consider "cmdline" the most universal API interface, and just appropriate for this type of function, regarding complexity of invocation and speed/overhead of calling it. Each language has a system() call, and shellscripts are most natural environment for cmdline execuatbles. No need for any dbus interface or whatever, for that wrapper API Sep 23 19:37:31 wotan147: I think you do that inside scratchbox Sep 23 19:38:19 but ask somebody who actually did it during last 24 months Sep 23 19:40:05 DocScrutinizer05 okay cause I mean in scratchbox i can not access to the documents of my host system Sep 23 19:41:23 wotan147: as DocScrutinizer05 said, the workflow is to log into scratchbox and do the packaging there Sep 23 19:43:07 shentey Sep 23 19:43:17 wotan147: and yes, you can't access files from your host system in scratchbox... you have to copy the files Sep 23 19:45:40 shenten okay so I copy everything from the host to guest then login and do the whole stuff? Sep 23 19:45:44 (assuming we're talking about scratchbox 1... I've no idea about sb2) Sep 23 19:46:07 wotan147: right Sep 23 19:56:03 either copy, or symlink. And of course you can access stuff *inside* SB from your host system, so your documents are maybe just in the wrong location from a host POV? Sep 23 19:56:25 err bindmount Sep 23 19:56:39 symlinks won't work Sep 23 19:56:54 hardlinks will Sep 23 19:57:06 or bindmounts Sep 23 19:57:48 which documents you're missing? Sep 23 19:58:26 all sourcecode and other files needed for packaging should already live inside SB Sep 23 20:01:16 I seem to remember I created normal dir structure and files inside scratchbox, and created a convenient symlink to that location in my homedir, like /home/jr/maemo/ACMEproject/ -> /scratchbox/home/user/home/whatever/user/bla/sources/home/user/ACMEproject/ Sep 23 20:03:03 symlinks from outside to inside SB work Sep 23 20:03:23 but you can't escape from SB to host, since it's a chroot environment Sep 23 20:06:16 shentey inside it says it doesnot find dh_make while debhelper is installed . do you have any idea? Sep 23 20:09:44 DocScrutinizer05 I was missing all the source directory since it was all on the host. I did sudo cp -r * /scratchbox....... and then chown from the host and now i can have it from inside the guest although I have problem with some files marked as root Sep 23 20:10:11 I will setup links as you said once the whole process work, now I have this weird think with dh_make Sep 23 20:11:05 wotan147: what command are you executing? Sep 23 20:12:50 shentey dh_make --createorig --single -e myemail@pro.com -c gpl but I have just spotted an internet that dh_make is not include with the debhelper of freemantel so apparently I have to dh_make from outside Sep 23 20:13:41 wotan147: alright... I'm not a .deb professional anyway ;) Sep 23 20:32:52 it now says rm: cannot remove `build-stamp': Permission denied when i dpkg-buildpackage do you know what it refer to? Sep 23 22:16:51 wotan147: I am guessing it is failing to remove the build-stamp found in the clean section of debian/rules Sep 23 22:19:46 you should copy the foopackage dir into scratchbox enviro and run "dpkg-buildpackage -sa -rfakeroot" in scratchbox as ARMEL on that folder. If the all the contains debian folder is ok it should build. Sep 23 22:20:33 yep that sounds correct Sep 23 22:20:43 It appears you are offline but hopefully someone will point you to the logs on your next appearance.... Sep 23 22:22:47 ;-) Sep 23 22:22:52 Even if it doesn't build you will get some sane error in scratchbox that you can work from, most probably debian/rules or debian/control file issues. Sep 23 23:22:14 any commands for checking if the N900 is connected to the Internet ? Sep 23 23:23:03 I've checked phone control - http://wiki.maemo.org/Phone_control#Networking Sep 23 23:23:27 but it has only enable, disable & state change callback Sep 23 23:23:34 not checking of current state Sep 23 23:31:50 $ awk 'BEGIN{f=1} ($2=="00000000" && $8 == "00000000"){f=0} END{exit f}' /proc/net/route && echo online Sep 23 23:34:22 r00t|home: well, that's indeed a possibility :) Sep 23 23:34:57 I expected something icd/icd2 related, but this looks like it would do too :) Sep 23 23:35:07 it's also more portable ;) Sep 23 23:35:14 r00t|home: BTW, is awk on the N900 by default ? Sep 23 23:35:19 no idea about the icd stuff Sep 23 23:35:31 awk is in better busyboxes Sep 23 23:36:36 r00t|home: OK, I think I'll go with this one :) Sep 23 23:36:51 Nokia-N900:~# dpkg -S $(which awk) Sep 23 23:36:51 busybox: /usr/bin/awk Sep 23 23:36:56 note lack of -power Sep 23 23:37:12 nice, that indeed looks like built-in Sep 23 23:37:21 also good point about portability Sep 23 23:37:59 well, awk should be probably available almost anywhere Sep 23 23:38:04 ofcourse it won't work if the device _believes_ is is online, but the connection is not actually working ;) Sep 23 23:38:18 r00t|home: hh :) Sep 23 23:38:48 also doesn't work in dial-on-demand setups Sep 23 23:38:59 sure, that might be an issue Sep 23 23:39:03 but i guess n900 handles that through some api calls, and not on routing level Sep 23 23:40:18 also, depending on use case it might be better parse the file in shell directly Sep 23 23:43:10 Nokia-N900:~# o= ; while read if dest gw f x x x mask x x x ;do [ "$dest" = 00000000 ] && [ $mask = 00000000 ] && o=1 ; done well, I need to run the check from Python Sep 23 23:43:57 THEN WRITE IT IN PYTHON DAMIT Sep 23 23:44:17 sure, good point :) Sep 24 00:23:20 r00t|home: seems to be working :) Sep 24 00:28:33 grrrr Sep 24 00:28:38 ~flashing Sep 24 00:28:39 rumour has it, maemo-flashing is http://wiki.maemo.org/Updating_the_tablet_firmware Sep 24 00:28:57 fsck libusb0.1.4 Sep 24 00:32:14 at least the wiki instructions work Sep 24 00:34:01 r00t|home: so thanks a lot ! :) Sep 24 00:34:14 r00t|home: BTW, this is how it looks like in the end: https://github.com/M4rtinK/modrana/blob/master/modules/device_modules/base_device_module.py#L229 Sep 24 01:05:56 r00t|home: wtf, what a cmdline?! Sep 24 01:06:56 [ "$o" ] && really? Sep 24 01:18:36 I just realized what friggin PITA it must be for linux-inept users to reflash their device Sep 24 01:19:25 DocScrutinizer05: that tests if $o is empty... Sep 24 01:19:39 or rather, not empty Sep 24 01:20:45 flasher-3.5 refused to work on my system due to 64bit libusb, then I had to blacklist phonet, then it refused to flash and I had to ^C and restart flasher to make it finally go thru ... after it failed halfway first tinme for unknown reasons Sep 24 01:21:12 r00t|home: I know shellscript a bit, but that's really obfuscated stuff Sep 24 01:21:16 $ help test Sep 24 01:21:18 -n STRING Sep 24 01:21:18 STRING True if string is not empty. Sep 24 01:21:38 you didn't test for -n Sep 24 01:21:45 see the second line Sep 24 01:21:48 you can omit the -n Sep 24 01:22:06 yeah, like I said, maximum obfuscated ;-D Sep 24 01:22:12 it's all rather natural to me... but then i've been doing bash for half a decade Sep 24 01:22:30 most people write utter nonsense like [ x$o = x ] Sep 24 01:22:58 has some rationale Sep 24 01:23:04 (you need to quote the variable, because during parsing the empty string is just skipped over otherwise, and you end up with [ ] which is a syntax error Sep 24 01:23:35 people who don't know quoting use the x=x variant Sep 24 01:23:45 (it still breaks if the variable contains spaces) Sep 24 01:24:11 well, I sometimes use x$A == xBLABLA Sep 24 01:24:29 use quotes instead and omit the x :P Sep 24 01:24:31 err s/==/=/ Sep 24 01:25:15 you can also use [[ instead of [ , which behaves much nicer, but it's only available in bash, not plain sh like busybox Sep 24 01:25:27 I think under certain circumstances even quoting with " fails Sep 24 01:26:05 e.g. when your $var is '"' Sep 24 01:26:10 quoting with " "fails" on exclamation points in interactive mode, because history exapnsion is for silly reasons allowed inside quotes Sep 24 01:26:16 else it always works Sep 24 01:26:43 some characters are still special inside ", like \ or $ Sep 24 01:27:01 but special chars in variable contents are no problem when the expansion is quoted Sep 24 01:27:25 yeah, I'm doing shellscripts since ~15 years now, though only occasionally Sep 24 01:27:45 i do way too much shell for unknown reasons... Sep 24 01:27:58 i guess the reason is it's always right at my fingertips when i have a shell open Sep 24 01:28:16 i liked that with basic back on my '64 Sep 24 01:29:03 you just start coding _right_there_, then copy it into a program/script file later Sep 24 01:29:25 funny enough, a colleague said "wowo, I never knew what you can do in shellscript" when reading http://maemo.cloud-7.de/maemo5/usr/local/sbin/bq27k-detail2 Sep 24 01:34:11 bit operations? Sep 24 01:34:27 i didn't know bash can do binary... only hex and octal :D Sep 24 01:35:07 hehe Sep 24 01:36:17 one thing bash can NOT do is store arbitary binary data in variables... at least 0-bytes break it... Sep 24 01:37:07 shouldn't quoting with ' fix all problems with special chars inside " ? Sep 24 01:37:24 dos1: sure, but it does not allow for variable expansion Sep 24 01:37:53 dos1: if you do it the "right" way, you're constantly switching quoting mode inside strings, which isn't beautiful either Sep 24 01:38:10 echo 'hello '"$world"'!' Sep 24 01:38:46 oh, i see Sep 24 01:38:57 in [ x$o = x ] context ' sure won't work Sep 24 01:39:51 but " works fine there Sep 24 01:40:01 dos1: I'm about to clean the device for Nikolaus Sep 24 01:40:37 or you use [[, which does not need quotes at all, because it's a shell syntax construct and not a command, so normal quoting rules do not apply Sep 24 01:40:47 add a FPC of the 3 pcs I received today, some GTA bare bones housing and GPS active antennae Sep 24 01:42:25 dos1: nikolaus maybe found a matching 1GB-RAM PoP, alas without NAND Sep 24 01:44:32 r00t|home: pun question: what type of command is ']' Sep 24 01:45:17 saturn:/var/log # which ] Sep 24 01:45:18 which: no ] in (/usr/lib64/mpi/gcc/openmpi/bin:/sbin:/usr/sbin:/usr/local/sbin:/root/bin:/usr/local/bin:/usr/bin:/bin:/usr/bin/X11:/usr/X11R6/bin:/usr/games:/opt/kde3/bin) Sep 24 01:45:26 DocScrutinizer05: nice! Sep 24 01:48:20 DocScrutinizer05: what are the adventages of running os from separate nand insead of emmc? bandwidth? Sep 24 01:50:53 ubifs on nand should be faster than ext3-on-emmc for small writes (common to OS use) Sep 24 01:51:33 With a modern kernel we could run f2fs instead, which performs about twice as fast as ext4 on emmc-type. Sep 24 01:54:38 Regardless, the more devices to spread IO traffic around to, the better Sep 24 01:55:21 :nod: Sep 24 01:56:57 dos1: no advantages Sep 24 01:57:35 but we need NAND for e.g CAL Sep 24 01:57:44 :-/ Sep 24 01:58:01 oh Sep 24 01:59:10 unless we build some massive cheat into kernel Sep 24 02:00:09 Considering the emmc doed about 1-4 WIOPS, if there's a write in-flight, the page-in of executable code from /opt can have to wait up to a second in worst case.. so I'd say it does have an impact Sep 24 02:00:16 it's not about those 256MB NAND and the 4 or 5 partitions on it, it's about kernel and userland using those Sep 24 02:01:27 (especially if sqlite-using things like.. well, conversations, browser, tracker is running) Sep 24 02:02:42 the new eMMC seem faster Sep 24 02:03:08 but still... missing on NAND is a pain Sep 24 02:03:28 sure we can boot from eMMC Sep 24 02:03:52 and maybe we even can cheat CAL and mtdoops Sep 24 02:04:29 but honestly I don't like that solution Sep 24 02:04:44 If it has a log-structured SLC cache, it will be able to do short bursts at 100-400 WIOPS.. if it just has faster raw nand inside though, it wont be a noticeable speed boost for multitasking os use Sep 24 02:04:45 I raher go for 512MB RAM and 512 NAND Sep 24 02:06:13 dman those 1GB RAM requests Sep 24 02:06:55 I'd rather take 1G/0G.. and put /home on uSD Sep 24 02:06:58 nobody offers 1GB RAM + NAND PoP MCP, except Nokia in N9 Sep 24 02:07:14 ShadowJK: no way Sep 24 02:07:24 that means you can't swap uSD anymore Sep 24 02:07:54 aside from that, why put /home on uSD? Sep 24 02:08:08 it's just fine on eMMC Sep 24 02:08:58 if that friggin DM3730 only had RAM bus on bottom chips Sep 24 02:09:29 I mean as a personal setup, to separate write intensive data from read intensive data Sep 24 02:10:00 that's your private non fremantle comaptibel setup Sep 24 02:10:08 so the write doesn't hold up reads Sep 24 02:10:17 not viable for Neo900 Sep 24 02:10:27 oh yes Sep 24 02:11:12 I already use swap on uSD for the same reason :P Sep 24 02:11:13 (and some other stuff, like moving gpodder's data files to uSD makes downloaading podcasts 5 times fasrer) Sep 24 02:11:44 so what do we need? 2 eMMC? Sep 24 02:13:34 How many mmc interfaces does omap have? :D Sep 24 02:13:37 I hoped for RAM in MCP eMMC chip, but it seems it has its own RAM bus, just simply sharing same chip package with eMMC Sep 24 02:14:22 (how many) to be evaluated if more than one can share same bus Sep 24 02:15:01 DM3730 has 2 separate IF iirc Sep 24 02:15:43 Bus sharing isn't very useful Sep 24 02:22:59 this would be awesome :D http://www.anandtech.com/show/7347/sandisk-announces-optimized-inand-extreme-emmc-for-bay-trail Sep 24 02:23:12 400 WIOPS sustained Sep 24 02:24:10 best of best I've seen so far has been 100-400 peak/burst Sep 24 02:24:37 i guess we can use iNand eMMC Sep 24 02:25:14 toldya there are faster eMMC nowadays Sep 24 02:45:56 I jst wish there was a flash (eMMC) with a smart controller and 1 or 2 GB of RAM "buffer" Sep 24 02:47:27 I thought those 32+1 combi eMMC would be this, but it seems they are not Sep 24 02:48:04 friggin usual mess of big business not even providing datasheets Sep 24 02:48:39 only pamphlets with marketing blabla Sep 24 02:51:19 what part are you trying to find datasheet for? Sep 24 02:51:38 mompls Sep 24 02:52:26 * jonwil reads backtrace Sep 24 02:52:34 chanlog Sep 24 02:53:07 32+1? Sep 24 02:53:08 you are talking about memory parts? Sep 24 02:53:10 http://talk.maemo.org/showthread.php?p=1372500#post1372500 Sep 24 02:55:40 KMK2U000VM Sep 24 02:57:51 >> Instead of interconnecting and using these memory chips separately within the device, MCP introduces the concept of unifying all the memory chips onto a single substrate. The chips are then connected on the substrate using one of several techniques, such as die-to-die bonding or die-to-frame bonding. Along with a substantial reduction in the board space used by the memory subsection, this technique significantly improves the Sep 24 02:57:52 performance of memory devices due to the tight coupling and shortest possible interconnections between the memory chips.<< **** ENDING LOGGING AT Tue Sep 24 02:59:59 2013