**** BEGIN LOGGING AT Tue Jan 24 03:00:00 2006 Jan 24 03:25:53 morning all Jan 24 03:26:56 hi Liam Jan 24 03:27:52 hi Marcin Jan 24 03:28:42 hows going tosa work? Jan 24 03:29:24 hrw|work: not really started yet. :( I've had a lot of support worki come in recently and it's really boged me down Jan 24 03:30:01 hrw|work: I should with any luck get something started next week. Jan 24 03:30:42 just asking ;) Jan 24 03:31:00 hrw|work: how your work going ? Jan 24 03:31:38 hi Liam Jan 24 03:31:48 hey Richard Jan 24 03:32:02 RP I'm booked into Novotel now Jan 24 03:32:20 lrg|home: quite well Jan 24 03:32:21 hi lrg Jan 24 03:32:29 hey pb_ Jan 24 03:32:54 lrg|home: Excellent. I'm quite looking forward to fosdem :) Jan 24 03:33:30 RP: same here, I'm arriving Thursday evening and meeting up with Dirk on the Friday Jan 24 03:33:53 lrg|home: I arrive midday on the Friday Jan 24 03:34:18 RP: excellent, we can all go out drinking on Friday night :) Jan 24 03:36:04 lrg|home: Sounds like a plan :) Jan 24 03:37:32 dont get too drunk Jan 24 03:38:43 heh, not me. Jan 24 03:39:30 hrw|work: My boss is likely to be around ;-) Jan 24 04:06:51 morning Jan 24 04:09:24 hey XorA Jan 24 04:15:15 The monotone update yesterday ended with 529 missing files. I can either use monotone drop or monotone revert --missing. Which one should I rather use? Jan 24 04:17:03 use revert Jan 24 04:17:12 morning, dyoung Jan 24 04:18:51 NAiL: Thank you. I will do that. Jan 24 04:19:45 morning Jan 24 04:20:10 hi CoreDump|home Jan 24 04:20:19 morning and l8er Jan 24 04:23:33 hi CoreDump|home, XorA Jan 24 04:25:00 ~curse libpcre for shitty utf-8 support Jan 24 04:25:03 May you be reincarnated as a Windows XP administrator, libpcre for shitty utf-8 support ! Jan 24 04:30:15 I want to try gtk on the framebuffer directly -- is there a gtk package available for this, or should I configure manually? Jan 24 04:33:14 koobla: I dont think there is a bb for this Jan 24 04:35:36 ok, then I will go ahead and hack what I got :) Jan 24 04:42:17 03coredump 07org.oe.dev * reffe6a45... 10/packages/altboot/ (altboot.bb files/altboot.func files/init.altboot): Jan 24 04:42:18 altboot: Jan 24 04:42:18 - Reworked moving old mountpoints into the new rootfs Jan 24 04:42:18 - Optimized pivot_root'ing Jan 24 04:43:14 CoreDump|home: does whole altboot should go into branch? Jan 24 04:43:41 hmm, it wouldn't hurt Jan 24 04:52:10 Once we have arm kexec functionality under 2.6, a lot of the altboot code might become unneccessary... Jan 24 04:53:13 RP: only on 2.6 machines.. Jan 24 04:54:38 hrw|work: Yes, collie would still need it Jan 24 04:54:46 RP: poodle, tosa too Jan 24 04:55:14 RP: kexec? what does it do? Jan 24 04:56:06 CoreDump|home: It loads a new kernel Jan 24 04:56:15 sweet Jan 24 04:56:40 CoreDump|home: I have a kexec patch a user provided to test... Jan 24 04:56:40 I don't see how that makes booting from other devices easier tho Jan 24 04:56:55 RP: still compiling .dev here Jan 24 04:56:57 CoreDump|home: kexec /media/card/zImage.bin Jan 24 04:57:12 huh Jan 24 04:57:27 CoreDump|home: Change the kernel commandline to say root=/dev/mmcblk0p1 Jan 24 04:57:29 probably silly question -> is anyone planning to be at fosdem ? Jan 24 04:57:29 you can change the kernel in a _running_ system? Jan 24 04:57:46 CoreDump|home: It launches a new kernel Jan 24 04:57:50 CoreDump|home: yes Jan 24 04:57:59 holy cow Jan 24 04:58:01 The new kernel replaced the old one in memory and runs... Jan 24 04:58:18 with the system beeing in which state? Jan 24 04:58:31 "newly booted" I assume? Jan 24 04:58:53 Ifaistos: yeah, I think numerous members of Team OE will be there Jan 24 04:59:04 ie: trash all running processes and run INIT again? Jan 24 04:59:19 CoreDump|home: Yes, it will do a complete init from scratch Jan 24 04:59:27 hmmm i was thinking of going also Jan 24 05:00:07 very nice but kinda overkill =) altboot could play lilo in that case. Select kernel on boot-up, set kernel CMDLINE for kexec etc Jan 24 05:00:38 * CoreDump|home shrugs we will see =) Jan 24 05:00:53 CoreDump|home: Yes. In some ways its overkill, in others its quite useful. Both types of operation have their uses :) Jan 24 05:01:01 indeed Jan 24 05:01:16 CoreDump|home: I just wanted you to be aware of it as it might make 2.6 support easier :) Jan 24 05:01:29 Ifaistos: überhackers mickeyl and zecke will be there, at least, as will I Jan 24 05:01:48 easier? lol It means users can dual-boot between 2.4, 2.6 and Sharp. That's brilliant Jan 24 05:02:22 * mithro_ is trying to find a cheap way to get from Rome to Brussels :P Jan 24 05:02:39 mithro: hey Jan 24 05:03:07 hey zecke Jan 24 05:03:14 Ifaistos: also others such as lrg|home, do13, most of openedhand and myself Jan 24 05:03:22 * mithro was violently ill last night :/ Jan 24 05:03:29 mithro: well the tunnel is up buit had it has no ip address associated Jan 24 05:03:43 I think i need to "squiz" fosdem in.... Jan 24 05:03:46 :) Jan 24 05:04:14 CoreDump|home: Its possible 2.6 is good enough on collie to let you boot an initial kernel to kexec something else. I know poodle 2.6 and tosa 2.6 are certainly capable of that Jan 24 05:04:15 zecke: the tunnel device on the host shouldn't have an address associated with it Jan 24 05:04:26 anyone giving presentation ? Jan 24 05:04:41 mithro: well /sbin/ifconfig eth0 up fails in the uml Jan 24 05:04:58 hrm, what does it say? Jan 24 05:05:01 Ifaistos: lrg|home on ASoC last I heard Jan 24 05:05:19 RP: that's where altboot fits in nicely. You can script up anything for a "normal" boot. Jan 24 05:06:00 CoreDump|home: I've never actually tried altboot :) Jan 24 05:06:11 TUNSETIFF failed, errno = 16 Jan 24 05:06:58 It replaces /sbin/INIT and offers you various boot-time options. Each option launches a script that can do basically anything you want. Jan 24 05:07:07 you tried restarting the linux? Jan 24 05:07:33 mithro: more than once Jan 24 05:08:17 hrm.... Jan 24 05:08:43 can you stop it for a sec? Jan 24 05:11:08 CoreDump|home: I'll have to experiment when I resurect the work I did on the ipaq hx2750 as it sounds useful there :) Jan 24 05:11:53 heh won't work on iPAQ, sorry. You sort-of need a keyboard to navigate the menu. It doesn't support arrow-keys Jan 24 05:13:00 CoreDump|home: Ah :) Jan 24 05:13:01 hrm, i'm looking at 173 EUR :/ Jan 24 05:17:33 why could a package be compiled, but not added to an image (if included in the image DEPENDS and RDEPENDS) Jan 24 05:19:29 katossi_uni: you need it in IPKG_INSTALL Jan 24 05:21:07 it is Jan 24 05:21:23 that's why I don't understand nothing Jan 24 05:21:24 welp, train is hela expensive Jan 24 05:23:53 * mithro ponders what to do... Jan 24 05:24:24 mithro: re Jan 24 05:24:24 does FOSDEM cost anything? I couldn't find any prices anywhere Jan 24 05:24:38 zecke: give it a go now Jan 24 05:24:56 looks better again Jan 24 05:25:15 don't resize your windows :P Jan 24 05:26:22 mithro: well... it runs in a screen and it didn't work among restarts Jan 24 05:27:07 ? Jan 24 05:27:19 so if you restart it now it won't work? Jan 24 05:27:54 mithro: it didn't work Jan 24 05:29:54 hrm... that sucks Jan 24 05:39:51 mithro: look for eurolines if you can stand to be in the bus for so long Jan 24 05:40:09 mithro: or search for budget flights Jan 24 05:40:43 currently searching for budget files Jan 24 05:40:52 mithro: maybe Ryan Air or a lookalike has something for you Jan 24 05:41:52 ryan air will cost me 160 Eur :/ Jan 24 05:42:18 virgin-express 176.35 EUR Jan 24 05:43:09 if i fly in on the 23rd it's a bit cheaper Jan 24 05:43:17 mithro: Eurolines is 127 EUR one way Jan 24 05:43:31 ouch Jan 24 05:43:37 mithro: but their price depend on time of stay, departure etc Jan 24 05:43:46 http://www.eurolines.com Jan 24 05:44:09 mithro: when I was still in school and didn't have any money, then I hitchhiked the autobahn. Absolutely cheap :-) Jan 24 05:44:31 It took me 4 DM from Frankfurt to Berlin. The 4 DM (2 EUR) was for eat+drink :-) Jan 24 05:44:49 schurig: did you walk? Jan 24 05:44:50 but it could take from 5 to 12 hours Jan 24 05:45:07 zecke: hitchhike ("trampen" auf deutsch) Jan 24 05:45:28 schurig: I know people that even earned some money on their way Jan 24 05:45:38 ("pih ficken" auf deutsch) Jan 24 05:45:41 poh even Jan 24 05:45:56 lol zecke Jan 24 05:46:12 If I fly in on the 23rd it'll cost me 107 EUR Jan 24 05:46:22 mithro: http://www.viavai.it/ Jan 24 05:46:27 (with ryan air) Jan 24 05:46:47 schurig: what is that? Jan 24 05:47:28 136.35 with virgin-express Jan 24 05:47:43 mithro: an organized car hitchhiking. Drivers register there that they go from A to B. And if you want to ride with one, you ask if they have something and pay a fee to them, then they get you the address/phone # of the driver Jan 24 05:48:39 i think my best bet is to try and fly in on the 23rd Jan 24 05:48:53 anyway I've got to run! Jan 24 05:48:54 see ya Jan 24 05:52:53 gm Jan 24 05:55:12 mehhh ERROR: dependency ${BOOTSTRAP_EXTRA_DEPENDS} (for task-hentges-bootstrap) not satisfied. Guess I _do_ have to rework that after all Jan 24 05:56:12 CoreDump|home: Just set BOOTSTRAP_EXTRA_DEPENDS="" Jan 24 05:56:17 or remove it Jan 24 05:57:53 yeah, BOOTSTRAP_EXTRA_DEPENDS ?= "" should do the trick if the VAR is emty beforehand right? Jan 24 05:58:05 CoreDump|home: yes Jan 24 05:58:08 ty Jan 24 05:58:56 CoreDump|home: but nothing sets it anymore (unless you have some local machine.conf files) and its honestly not needed if you inherit the image bbclass Jan 24 05:59:56 my problem is that my .bb's are in a CVS and they have to be usable on branch =) And I'd rather _not_ make a branch in my own CVS heh Jan 24 06:00:22 NOTE: package task-hentges-bootstrap-1.0: completed Jan 24 06:00:26 =) Jan 24 06:00:28 CoreDump|home: Fair enough :) Jan 24 06:01:31 I'm going to flash a .dev opie-image now. Let's see how 2.6 holds up Jan 24 06:01:49 * RP hides Jan 24 06:02:16 heh nahh Jan 24 06:02:45 anything I should know about? Any noteworthy bugs / problems? Jan 24 06:03:22 CoreDump|home: This is akita? There shouldn't be anything major missing. Perhaps just some polish Jan 24 06:03:42 I'm not sure of the status of akita's fstab... Jan 24 06:03:43 great =) Jan 24 06:04:26 * CoreDump|home is great at polishing bootstrap stuff Jan 24 06:05:53 flashing away Jan 24 06:06:16 CoreDump|home: did you noticed something more to polish recently? Jan 24 06:07:17 hrw|work: umm maybe turn off the logmessages from /etc/init.d/opie. Don't remember anything beside that ATM Jan 24 06:07:27 and well, Jan 24 06:07:50 GPE's bootsplash looks sweet, but INIT5 make a mess of it. Looks terrible Jan 24 06:08:04 gpe bootsplash + elpp == argh Jan 24 06:08:07 maybe one could do a "clear" in S00_blahh in rc5.d Jan 24 06:08:16 ~lart qpdf2 Jan 24 06:08:17 * ibot stuffs qpdf2 into a shiny new tin can and vacuum seals it Jan 24 06:08:35 17M archive with all .o etc files Jan 24 06:08:50 * CoreDump|home is getting _tons_ of "Empty flash at blahhh ends at blahh" messages Jan 24 06:09:10 CoreDump|home: They're "normal" for akita. Jan 24 06:09:24 I hope they do not indicate a problem? Jan 24 06:09:25 CoreDump|home: you need DISTRO_TYPE="release" then Jan 24 06:10:09 hrw|work: nahh, I want to see kmesg right now =) It's a .dev image after all Jan 24 06:10:10 CoreDump|home: No problem I'm aware of Jan 24 06:10:40 RP: ok. It would suck to have the flash killed due to some annyoing bug lol Jan 24 06:11:18 hmm Opie apears to run faster w/ 2.6 Jan 24 06:12:55 CoreDump|home: The flash in akita is different to any previously in Zaurus models. I borrowed my boss' device to write the driver and those warning can be safely ignored as far as I can see. Its mostly a standard mtd nand driver so if there is a bug, its in the mainline mtd code ;-) Jan 24 06:13:09 CoreDump|home: That'll be PREEMPT Jan 24 06:13:13 RP: =) Jan 24 06:15:18 the delay for "key-repeat" should be raised Jan 24 06:16:10 CoreDump|home: find best value with kbd_rate so it will be raised Jan 24 06:16:19 ok Jan 24 06:16:38 CoreDump|home: known issue. You can use console-tools to change it but a kernel patch would be the best solution Jan 24 06:16:50 right Jan 24 06:16:59 looking very good Jan 24 06:17:02 (a kernel patch to allow keyboard drivers to set their own keyboard repeat rates) Jan 24 06:17:05 CF auto-mounts Jan 24 06:17:21 SD doesn't auto-mount but can be mounted menually Jan 24 06:17:32 CoreDump|home: SD should automount too Jan 24 06:17:48 CoreDump|home: Probably the akita fstab Jan 24 06:18:13 right Jan 24 06:18:23 its still using mmcda* Jan 24 06:18:58 CoreDump|home: See the spitz one for a suitable 2.6 entry Jan 24 06:18:59 udev can generate symlinks for mmcda anyway... Jan 24 06:19:02 no its not lol Jan 24 06:19:41 hrw|work: I'd like to think long and hard before we made it do that though... Jan 24 06:20:36 I would prefer no not doing it Jan 24 06:21:00 hrw|work: me too. Especially if we split the machine types into 2.4 and 2.6 versions... Jan 24 06:22:05 RP: fstab is fine. The partitioning of my card was at fault. 2.6 doesn't ship with the small script that auto-detects the partition Jan 24 06:22:56 CoreDump|home: 2.6 uses a udev automount script which could probably get adjusted to work. I was also looking into pmount Jan 24 06:23:42 CoreDump|home: My todo list includes this, unless someone beats me too it :) Jan 24 06:23:54 =) Jan 24 06:24:16 is http://bugs.openembedded.org/ down? Jan 24 06:24:40 it is Jan 24 06:24:48 CoreDump|home: There's a bit of rewriting to be done for the move to pcmciautils (network devce handling) which will be done from scripts via udev as well Jan 24 06:26:32 RP: what exactly needs adjusting? My CF WLAN card just went online w/p problems Jan 24 06:26:43 CoreDump|home: At the moment it uses pcmcia-cs Jan 24 06:27:08 and that will be replaced with pcmcia-utils? Jan 24 06:27:18 CoreDump|home: yes Jan 24 06:27:25 I'll have a look Jan 24 06:27:56 * RP -> lunch Jan 24 06:28:24 tries to set up USB networking Jan 24 06:33:13 CoreDump|home: http://hrw.one.pl/tmp/usb-gadget.sh - use opie-sh Jan 24 06:38:32 CoreDump|home: if you can test it under opie/2.6 then it would be great Jan 24 06:38:51 I'm looking at it now Jan 24 06:39:18 opie-sh still needs to compile here tho Jan 24 06:40:32 it does not do any error checking Jan 24 06:41:19 zecke, you about? Jan 24 06:41:38 yes Jan 24 06:41:53 I assume holger should have r/w access to subverion repo? Jan 24 06:42:02 schurig? Jan 24 06:42:05 yeah Jan 24 06:42:11 yes Jan 24 06:42:16 he created an account Jan 24 06:42:17 ok Jan 24 06:42:18 zecke: yes? Jan 24 06:42:26 schurig: you want to have write access? Jan 24 06:42:37 zecke: I haven't send the mail yet, but yes. Jan 24 06:42:38 schurig: I'm still svk smerging the repository Jan 24 06:43:07 Crofton: your repository is not setup in the usual SVN way, with trunk/ and branches/. Doing branches the SVN way is therefore not possible. Jan 24 06:43:15 hmmm Jan 24 06:43:22 that is zecke issue :) Jan 24 06:43:32 zecke: your repository is not setup in the usual SVN way, with trunk/ and branches/. Doing branches the SVN way is therefore not possible. Jan 24 06:43:33 :-) Jan 24 06:43:38 he should be able to move things around when he is done building it Jan 24 06:44:00 schurig: heh? Jan 24 06:44:15 schurig: trunk/ branches/ tags/ are just directories Jan 24 06:44:24 svn mkdir /branches Jan 24 06:44:36 svn cp org.oe.dev /branches/schurig-branch Jan 24 06:44:45 zecke: right. But all the source is in the root. So mkdir trunk/ and move all into it Jan 24 06:45:11 zecke: otherwise, when you have 10 branches and you do the intial checkout as descripted, you get all 10 branches as well. That's not common. Jan 24 06:45:33 schurig: well I could stop the merging Jan 24 06:45:46 schurig: wait Jan 24 06:45:47 just move things about later Jan 24 06:45:50 zecke: http://svnbook.red-bean.com/nightly/en/svn.branchmerge.using.html Jan 24 06:45:53 schurig: we have /org.openembedded.dev Jan 24 06:46:10 schurig: I know svn cp, not even the svn main server uses trunk/ tags/... Jan 24 06:46:43 zecke: we should have org.openembedded.dev/trunk/conf/bitbake.conf (for example). And then later org.openembedded.dev/branches/dev.slug/conf/bitbake.conf if one makes this branch. Jan 24 06:47:00 schurig: whatever (besser wisser)... Jan 24 06:47:04 and in org.openembedded.dev/ there should not be any other file/directory besides the branch/ and trunk/ directory Jan 24 06:47:06 schurig: you know that svn cp is O(1)? Jan 24 06:47:19 schurig: we could even create your 'schurig' view Jan 24 06:47:28 schurig: so the dirs get ordered the way you want... Jan 24 06:47:56 schurig, you should have write access now Jan 24 06:48:03 schurig: I do not want trunk/ branch/ tags/ under org.oe.dev Jan 24 06:48:11 schurig: this wouldn't match what we do now... Jan 24 06:48:21 schurig: .dev is the development branch... Jan 24 06:48:36 schurig: what you could call trunk if you like to HEAD, or foo, or even org.oe.dev ;) Jan 24 06:48:39 zecke: this has nothing to do with besserwisser or O(1). When you have "normal" (HEAD) files and directories in org.openembedded.dev/ AND you have the org.openembedded.dev/branches/ directories, then you can't just do an initial checkout of HEAD, you'll get all branches/ and all tags/. That's highly inconvenient. Jan 24 06:48:55 zecke: if you evaluate an SCM tool, then you should stick to it's customary use. Jan 24 06:49:07 zecke: look at the pictures in http://svnbook.red-bean.com/nightly/en/svn.branchmerge.using.html Jan 24 06:49:09 schurig: well svn co http://foo/org.openembedded.dev . Jan 24 06:49:23 schurig: I may place dirs where I want to... Jan 24 06:49:29 zecke: and where do you put tags in? Jan 24 06:49:30 schurig: org.openembedded.dev is trunk Jan 24 06:49:58 schurig: OE's development doesn't fit too well into your branch/ tags/ trunk kind of thinking Jan 24 06:50:01 zecke: in https://ossie-dev.mprg.org/repos/oe/tags? Jan 24 06:50:29 zecke: *IF* we ever switch (assuming people vote for it), then it might be nice to have tags, wouldn't it? Jan 24 06:50:44 and branches in https://ossie-dev.mprg.org/repos/oe/? Jan 24 06:50:47 schurig: wtf... I just smerging all 4000 revs Jan 24 06:50:55 schurig: look at the current rev number... Jan 24 06:51:04 that would work, but it's quite unusual. Most SVN repositories are laid out different. Jan 24 06:51:04 schurig: don't annoy me now, please don't do it Jan 24 06:51:21 schurig: the trunk/ branch/ tags/ thinking does not manch OE style of development Jan 24 06:51:25 zecke: don't get so easily annoyed Jan 24 06:51:37 I would recommend patience Jan 24 06:51:46 schurig: is nslu, dreambox just a OE branch? Jan 24 06:51:54 most of the svn organization is social, not required Jan 24 06:52:01 schurig: and is the main OE development in /trunk? I think not Jan 24 06:52:31 schurig: we would have official/org.oe.dev official/org.nslu-linux official/de.mn... Jan 24 06:52:43 schurig: as these are all official HEADs for their development Jan 24 06:52:58 schurig: well you could place these dir under /trunk Jan 24 06:53:05 CoreDump|home: One thing to note - usb client doesn't work with windows on the pxa270 (yet) Jan 24 06:53:31 schurig: one point of the trial is to experiment, coming with the bible and saying that a svn tree needs to be /trunk /tags /branches... Jan 24 06:53:34 RP: thank good I don't do windows heh heh Jan 24 06:53:41 *shakes head* Jan 24 06:54:08 you have 1200 revs to go before you have to solve this :) Jan 24 06:54:18 zecke: everything in branch/ is also a head. Yes, you're right, it boils down to the view if they are all equal brothers or if one is master of the universe and the other siblings. Jan 24 06:54:21 CoreDump|home: I have patches and some clues on how to make it work, next time I have some time to spend on it... Jan 24 06:54:34 CoreDump|home: My todo list is unmanagable atm :-/ Jan 24 06:54:41 =) Jan 24 06:55:08 zecke: however, even in monotone I thought that org.openembedded.dev is the master, the head where everything culminates. Forgive if I was wrong :-) Jan 24 06:55:09 schurig: at first I synced all org.oe.dev dirs to / which was bad Jan 24 06:55:20 schurig: with /org.oe.dev we have all freedom to organize... Jan 24 06:55:36 schurig: with distributed scm's you don't have a master Jan 24 06:55:56 schurig: it is social agreement as well. The main OE folks push to org.oe.dev Jan 24 06:56:05 schurig: the nslu people to org.nslu... Jan 24 06:56:11 zecke: wrong. Even with Bitbake we had a master. The master is the thing that is documented in our Wiki where people make their initial checkout. Jan 24 06:56:12 schurig: and once in a while we synchronized Jan 24 06:56:27 schurig: Bitbake? Jan 24 06:56:29 zecke: if I commit something and it doesn't and in this master, then it is not "world-wide" published. Jan 24 06:56:41 the key thing here is how good svn/svk is at synching across different branches Jan 24 06:57:02 which is a little different from how I think most people use svn Jan 24 06:57:04 zecke: so even our way of a distributed repository has a master, by way of the documentation (and also by the way we make snapshots and publish them on the website) Jan 24 06:57:25 schurig: well switch to the nslu perspective Jan 24 06:57:54 schurig: occasionally they get the bugs from OE Jan 24 06:57:59 schurig: and in their wiki... Jan 24 06:58:29 schurig: we have three subprojects with active development Jan 24 06:58:39 schurig: they are all official, they would be all in 'trunk' Jan 24 06:58:53 schurig: now each of them can go into a stablisation phase Jan 24 06:58:54 zecke: and, if nslu2 people modify, say, samba, do the merge that back in org.openembedded.dev ? Jan 24 06:59:07 schurig: this wehre one could use the branches Jan 24 06:59:14 schurig: now people can create personal branches as well Jan 24 06:59:51 zecke: you can use branch/nslu2 for stabilization, too. This is what happens in KDE's SVN repository, e.g. with branches/KDE/3.x/ Jan 24 07:01:16 schurig: I think we agree on that Jan 24 07:01:34 org.oe.dev/{trunk,tags,branches} looks wrong to me Jan 24 07:01:53 /trunk/org.oe as well Jan 24 07:02:05 this is why I decided to postpone discussion about the structure to later Jan 24 07:02:36 zecke: we just disagree on the point if we have a "Prima inter pares" or not. Once this is settled (if ever) the resulting directory structure is logical Jan 24 07:03:01 is a logical result of the decision Jan 24 07:03:12 schurig: wait another two days after I have replayed all revs... Jan 24 07:03:26 schurig: then we can do tree copying Jan 24 07:03:32 horse cart :) Jan 24 07:04:32 Crofton: I don't get that :) Jan 24 07:04:47 Crofton: me neither Jan 24 07:04:51 putting the cart before the horse Jan 24 07:05:16 my way of saying things have to be done in order :) Jan 24 07:05:19 Crofton: Is that the meaning of "the tail wiggles with the dog" ??? Jan 24 07:05:26 something like that Jan 24 07:05:36 the horse pulls thcart, it can't push the cart Jan 24 07:05:53 Crofton: I'm still not sure how I can apply this to the current situation Jan 24 07:06:14 well, you have to complete the repo, before you can try different strcutures Jan 24 07:06:42 and since cp's and mv's are lightweight in svn, plenty of arrangements may be tried Jan 24 07:07:04 zecke: I came acros this branch issue because I wanted to convert http://oe.handhelds.org/cgi-bin/moin.cgi/SvnScmTrial in a way so that we have two parts in it, an "SVN Phrasebook" and an "SVK Phrasebook". Now I won't do this, until I see what you have in petto. Jan 24 07:08:41 schurig: what stops you? Jan 24 07:08:54 schurig: you can just create tags/ and branches and trunk/ now if you want to Jan 24 07:09:10 schurig: and after the smerge you can svn mv it... Jan 24 07:09:19 zecke: WHERE should I create them? Jan 24 07:09:37 schurig: oe/trunk oe/tags/ Jan 24 07:09:45 schurig: we now have oe/org.openembedded.dev Jan 24 07:09:50 schurig: so with svn mkdir... Jan 24 07:10:25 zecke: ok, then I have to svn mv all things around. But if I got you right, then you want to make changes in this SVN repository during the next two days? Or maybe I didn't get you right? Jan 24 07:10:57 zecke: also, you just said to me that https://ossie-dev.mprg.org/repos/oe/org.openembedded.dev/trunk/ and https://ossie-dev.mprg.org/repos/oe/org.openembedded.dev/branches/ doesn't sound right to you. If other agree, then maybe my idea is nonsense after all. Jan 24 07:11:12 schurig: one dir? Jan 24 07:11:18 in this case maybe a https://ossie-dev.mprg.org/repos/oe/nslu2.dev/ dir should be made Jan 24 07:11:19 all things == one dir Jan 24 07:11:40 schurig: then we can move around once more Jan 24 07:11:44 and as long as this isn't really known, there's no point in documenting this and writing a phrasebook Jan 24 07:11:46 schurig: this is the idea of the trial Jan 24 07:11:56 schurig: we will throw it away afterwards.. Jan 24 07:13:27 ~seen RP Jan 24 07:13:32 rp is currently on #openzaurus #gpe #handhelds.org #oe. Has said a total of 1231 messages. Is idling for 18m 58s, last said: 'CoreDump|home: My todo list is unmanagable atm :-/'. Jan 24 07:13:49 hi schurig Jan 24 07:14:05 busybox's route is beginning to piss me off Jan 24 07:14:12 RP: I'm using the pxa-cpufreq patch from your repository and found a fix for it Jan 24 07:14:46 RP: in drivers/pcmcia/pxa2xx_base.c line 144 you have a warning, wondering if this is right Jan 24 07:15:27 schurig: That warning has always been there - I honestly don't know if it applies or not Jan 24 07:15:31 RP: if you have the powersave governor (with the 4 PXA frequencies), then the memory clock is always 99532, in this case you woudn't need this Jan 24 07:15:32 schurig: Fixes are welcome Jan 24 07:15:57 RP: bit if you use performance, you'll have 6 frequencies, and the memory clock varies. Jan 24 07:16:17 03rwhitby 07org.oe.dev * r2e1b97f9... 10/ (4 files in 2 dirs): ftpd-topfield: Updated to 0.6.3 Jan 24 07:16:30 RP: so the #warning can go away. However, the fix now is pxa2xx_pcmcia_set_mcxx(skt, freqs->new) is wrong, it needs to be pxa2xx_pcmcia_set_mcxx(skt, freqs->new /10) Jan 24 07:16:49 If I want to set EXTRA_OECONF for all arm-arches, is there a simple way to do it? (both be and le) Jan 24 07:16:50 otherwise MCIOx get's programmed so slow that you get DMA starvation, visible on the PXA framebuffer Jan 24 07:17:20 the same some lines below, it must be pxa2xx_pcmcia_set_mcxx(skt, freqs->new / 10) Jan 24 07:17:39 schurig: That looks like a good find, thanks. I'll add that to the patch Jan 24 07:18:17 schurig: I didn't write the patch and haven't looked too closely at it yet but hopefully once I do. I'll probably try for mainline with it... Jan 24 07:18:40 RP: I'm using this patch since some weeks, with a userspace app that changes frequency Jan 24 07:19:06 RP: to gether with core voltage modification via a callback I can save a substantial amount of mW :-) Jan 24 07:19:08 morning Jan 24 07:19:54 schurig: Is your core voltage changing patch suitable for other machines? I might be interested in that... Jan 24 07:20:19 schurig: I assume you didn't find any other problems with the cpufreq patch? Jan 24 07:20:34 RP: of course not. You need a GPIO somewhere (e.g. PXA, CPLD, AC97 codec) and this is highly machine specific. Jan 24 07:20:56 one of our hardware can only select between two voltages, the other between 4 voltages ... Jan 24 07:21:13 schurig: I mean is it clean enough to be easily altered for other hardware :) Jan 24 07:21:35 RP: moment, I post this to pastebin.ca ... Jan 24 07:22:05 http://pastebin.ca/38353 Jan 24 07:22:32 03nail 07org.oe.dev * r8de532d6... 10/conf/distro/ (debianslug.conf openslug.conf): debianslug/openslug.conf: Fix breakage caused by _EXTRA_RDEPENDS. Jan 24 07:22:35 RP: so, it's quite simple Jan 24 07:22:36 03nail 07org.oe.dev * r50d95cd7... 10/packages/meta/slugos-packages.bb: slugos-packages: Remove memtester from uclibc packages Jan 24 07:23:27 RP: oh, and you need a "cpufreq_register_notifier(&rt3000_cpufreq_notifier_block, CPUFREQ_TRANSITION_NOTIFIER);" somewhere Jan 24 07:23:45 hey schurig, been a while Jan 24 07:24:03 anyone kick openembedded.org yet? Jan 24 07:24:16 schurig: It is a little simpler than I was expecting :). The zaurii have an i2c voltage controller which I was thinking this might be similar to Jan 24 07:24:49 ooo, i2c my friend Jan 24 07:24:55 emte: yes. I'm seldom in IRC nowadays. Jan 24 07:25:24 emte: and days after the monotone switched I stopped contributing to OE (it was soooo slow) Jan 24 07:25:42 yeah :( Jan 24 07:25:46 emte: now if OE switches away from OE I have a bunch of things to contribute Jan 24 07:26:04 schurig: please watch the scm issue and vote :) Jan 24 07:26:05 switches away from MT, i meant Jan 24 07:26:06 schurig, you didnt try using kergoth's svk version? Jan 24 07:26:51 emte: I just saw that there is a vote today, when I opened akregator (where I saw the RSS feed of the OE wiki). I haven't made much besides discussing with zecke about the SVN directory layout. Jan 24 07:27:04 hg has alomost finished most of the issues that were missing before :) Jan 24 07:27:06 emte: I actually want to test pure SVN, SVN+SVK and GIT. Jan 24 07:27:35 i only read all the bad things about GIT in #mercurial Jan 24 07:27:36 emte: months ago I tested "hg" and liked it, but at that time there was no good support for branches Jan 24 07:27:46 emte: hehe :-) Jan 24 07:28:16 they have lots of good things to say about monotone, cept the speed issue Jan 24 07:28:34 which apperently is due for another huge upgrade Jan 24 07:29:35 :( openembedded is still down Jan 24 07:31:32 so the WCF54G works with the wavelan_cs driver and macchanger works great Jan 24 07:32:00 maybe you only need Agere for the G part of the chip ? Jan 24 07:32:35 or the Agere cards were begore Cisco and Linksys merged Jan 24 07:32:44 would be my other guess Jan 24 07:35:11 wavelan_cs is indeed the driver for Agere wavelan cards. Jan 24 07:36:11 ... Jan 24 07:36:17 there is also this BADBADBAD agere driver wl_lkm_718_release.tar.gz, there the kernel module is named wlags49_h2_cs.ko Jan 24 07:36:23 or wlags49_h1_cs.ko Jan 24 07:36:48 took me a while to get this working with 2.6.13 and then 2.6.15, but it's the only driver the the CF cards I have here. Jan 24 07:36:59 then why was everyone pointing to wl_lkm? and Agere not listed in wavelan description? Jan 24 07:37:29 totally different written to all other Linux drivers, e.g. one code base for NDIS2, NDIS3, Novell, DOS, WinCE, RTOS and whatever OS. Thousands of #ifdef's ... Jan 24 07:37:50 but it works and it works with WPA even, so I'm using it Jan 24 07:37:56 heh Jan 24 07:38:21 ah Jan 24 07:38:38 my issue was the "adjustable MAC" Jan 24 07:38:51 after i figured out wavelan_cs worked Jan 24 07:39:41 emte: is wavelan_cs part of the 2.6 kernel? Jan 24 07:40:03 i am using the 2.4.18 one in familiar-0.8.3 Jan 24 07:40:24 emte: I'm using 2.6.15 :-) Jan 24 07:40:38 die, old kernels, die Jan 24 07:40:51 i did come across a few documents that also said it was a Prism3 card ... Jan 24 07:41:17 linksys/cisco was vague on the issue Jan 24 07:41:46 Ah, I have a Hermes 2 card Jan 24 07:42:06 emte: because you spoke about Agere I thought you have a Hermes card, too. Jan 24 07:42:44 nope, just the WCF54G Jan 24 07:43:20 schurig, you should should try the macchanger bb i posted to the mailing list Jan 24 07:43:27 its fun Jan 24 07:46:17 anyway off to class Jan 24 07:50:29 03nail 07org.oe.dev * ref393c0f... 10/packages/lvm2/lvm2_2.01.15.bb: LVM2: Upstream upgrade to 2.01.15, fix o_direct bug on arm Jan 24 07:50:33 03nail 07org.oe.dev * re3e4a744... 10/packages/device-mapper/ (2 files in 2 dirs): device-mapper: Upstream upgrade to 1.01.05 Jan 24 07:50:38 03nail 07org.oe.dev * r86a447fd... 10/ (2 files in 2 dirs): Add device-mapper/lvm2 to slugos feed Jan 24 08:07:04 i am trying to build a specific package... how should the filename be formed ? bitbake quild-native and bitbake quild-native-0.39 and bitbake quild-native-0.39.bb fail with a "nothing provides dependancy" message Jan 24 08:08:19 zecke, I may have a setup problem with the svn server Jan 24 08:08:34 I am starting to test my stuff on it and it is reidiculously slow Jan 24 08:08:56 Crofton: oh :} Jan 24 08:09:16 Ifaistos: PN_PV.bb Jan 24 08:09:41 Ifaistos: It would be bitbake quilt-native or bitbake -b /somepath/somebb_someversion.bb Jan 24 08:12:03 RP: what I wrote about just doing clk/10 in pxa2xx_pcmcia_frequency_change is nonsense Jan 24 08:12:54 RP: if old and new would be the memclock values, it would work, but they are cpuclock values, so it's utterly wrong Jan 24 08:14:08 zecke, I can browse my repo fine, but command line stuff isn't working well Jan 24 08:14:17 are you seeing anything similar? Jan 24 08:14:38 I'm doing a ci atm.. so it works here Jan 24 08:14:49 does it seem really slow? Jan 24 08:15:09 let me know when you are idle, maybe ci'sslow things done??? Jan 24 08:15:42 schurig: Ah, I can see that being a problem Jan 24 08:15:54 schurig: Do you use pxafb btw? Jan 24 08:17:10 RP: Thanks ! I was trying the PB_PV.bb without the -b and full path.... Jan 24 08:17:31 Ifaistos: Keep in mind that -b doesn't resolve dependencies Jan 24 08:17:34 RP: ok, assume the CPU is fast (398131 kHz). When you go to very slow (99532 kHz) it would work, because 99532/10 is 9953 and this a valid parameter for pxa2xx_pcmcia_set_mcxx(). Jan 24 08:18:01 In function pxa2xx_pcmcia_set_timing() you can see that pxa2xx_pcmcia_set_mcxx() expectes memclock in kHz, not cpuclock Jan 24 08:18:25 is there a way to build a specific version package and resolve them ? Jan 24 08:18:43 so if you go from slow to fast, then calling pxa2xx_pcmcia_set_mcxx() with 39813 is wrong, this programs very weird timings in MCMEMx, MCATTx, and MCIOx Jan 24 08:19:06 Ifaistos: Set PREFERRED_VERSION_somepackage = "xxx" in local.conf and bitbake somepackage Jan 24 08:19:43 schurig: So we need a way to work out the memoryclock. Is there not a function in mach-pxa/generic.c for this? Jan 24 08:20:12 ok. i got it. bitbake always checks local.conf to get it's settings regardless of a single file or an image be build.... Jan 24 08:21:00 RP: in arch/arm/mach-pxa/cpu-pxa25x.c are tables with the frequencies, I don't know if the needed function is yet there Jan 24 08:23:00 schurig: mach-pxa/pxa25x.c has a get_memclk_frequency_10khz function Jan 24 08:24:01 RP: yes, this function is used in pxa2xx_pcmcia_set_timing(). But this way you can just get the CURRENT memclock frequency, not the old/new in the switch Jan 24 08:25:16 schurig: ok. I wasn't 100% which numbers you needed Jan 24 08:38:12 it was suggested that I join this channel to ask the following question. I am looking for a cheap, small, quiet, low heat, fanless pc that I can put fedora on. I am thinking that it would only need a serial port, 2 ethernet ports and a compact flash slot (instead of a HD) no video or sound. Does anyone here know of such an animal? Jan 24 08:38:49 MrPotter: probably, not the best room for that - jump on over to #elinux, and I'll respond to you Jan 24 08:39:17 ok, sorry, will do Jan 24 09:38:20 zecke, false alarm on svn perf issue at my end Jan 24 09:38:26 cockpit error. Jan 24 09:49:47 Crofton: hehe, but it feels slower now Jan 24 09:49:55 heh Jan 24 09:49:59 shouldn't be Jan 24 09:50:18 the machine is behind a small router Jan 24 09:50:44 and it always act slow/werid when I hit the machine via the router from inside the network Jan 24 09:51:00 I thought I was on my home machine in that window, thus the confusion Jan 24 09:56:38 RP: pcmciautils just cause a kernel oops Jan 24 10:01:14 CoreDump|home: Did you remove pcmcia-cs Jan 24 10:01:28 and rebooted, yes Jan 24 10:01:31 CoreDump|home: Its always worked ok for me :-/ Jan 24 10:01:50 I removed an unmounted card from the slot -> oops Jan 24 10:02:19 [] (sys_umount+0x0/0x24c) from [] (sys_oldumount+0x14/0x18) Jan 24 10:02:19 [] (sys_oldumount+0x0/0x18) from [] (ret_fast_syscall+0x0/0x2c) Jan 24 10:02:19 note: umount[1932] exited with preempt_count 2 Jan 24 10:02:30 lets few lines Jan 24 10:04:22 CoreDump|home: Ah, removing mounted cards whilst in the mount point in a shell or similar does oops Jan 24 10:04:33 CoreDump|home: Fixing this is painful Jan 24 10:04:41 it wasn't mounted Jan 24 10:04:52 in fact, "mount" freezes when I try to do so heh Jan 24 10:08:14 CoreDump|home: Can you pastebin the full oops? Jan 24 10:08:29 err, to late =) Jan 24 10:08:32 CoreDump|home: Is sounds like it was mounted as what else would it be umounting? Jan 24 10:08:40 lets see if I can make it oops again Jan 24 10:09:02 btw ooopes.. I finally get max3387 so will build serial cable soon ;) Jan 24 10:09:16 RP: it wasn't mounted. It was inserted during boot-up but _not_ auto-mounted. Jan 24 10:10:05 CoreDump|home: It had to be umounting something... Jan 24 10:10:52 trying again now. Inserted card before power on Jan 24 10:12:00 cu Jan 24 10:13:15 no oops now sigh Jan 24 10:13:57 and it's beeing auto-mounted after I created the device file Jan 24 10:14:40 "created the device file"? with udev running? Jan 24 10:15:12 I just checked again, hda1 was not created on the first boot. Jan 24 10:15:17 now it is Jan 24 10:15:25 It sometimes doesn't see cards if they inserted when booting. Remove and insert and it does Jan 24 10:15:39 CoreDump|home: How does it compare to 2.4 then? :) Jan 24 10:16:00 heh 2.6 is _fast_ Jan 24 10:16:19 didn't encounter any hardware related problems yet. everything appears to work well Jan 24 10:16:39 ok, screen flickering in aside after bl->on =) Jan 24 10:17:42 pb_ : is there a fixed point integer cairo implementation available? Jan 24 10:17:56 We give a lot more bl options than 2.4. Some are stlightly unstable... Jan 24 10:22:56 hmm WLAN card is not configured Jan 24 10:23:11 zecke: I don't think so. Jan 24 10:23:14 CoreDump|home: This is due to the missing pcmcia-cs scripts Jan 24 10:23:30 CoreDump|home: Which is the reason we've not switched to pcmciautils yet Jan 24 10:23:36 well, that gives me something to do then =) Jan 24 10:24:17 CoreDump|home: You'll want to convert them to use udev - see the automount udev script I wrote as an example Jan 24 10:24:52 CoreDump|home: Eventually, I plan to have these scripts as part of a device manager daemon type package Jan 24 10:25:38 pb_: so Gtk+ apps got slower on fpu less systems (using hard float) Jan 24 10:25:39 that would be neat Jan 24 10:25:43 pb_: was that noticable? Jan 24 10:26:54 zecke: It was noticable Jan 24 10:27:18 is that on some ones agenda to fix? Jan 24 10:27:39 zecke: not on anyones I know Jan 24 10:28:40 http://www.gtkmm.org/docs/gtkmm-2.4/docs/tutorial/html/ch14s07.html <- that looks nice Jan 24 10:28:51 zecke: They also got slower on softfloat due to increased numbers of floating point ops Jan 24 10:28:52 but without a fast cairo... Jan 24 10:29:56 zecke: not sure, I haven't actually tried a Cairo-enabled gtk on my ipaq yet Jan 24 10:30:16 I imagine it might be noticeable, yeah Jan 24 10:31:26 This is a good day. My first oe image for the epia is up and running :) Jan 24 10:31:47 and it only took 4 days to master ;) Jan 24 10:32:16 heh Jan 24 10:32:19 rock Jan 24 10:35:24 now in order to "customize" the image i sould add packages in the /conf/distro/familiar for example or start a new distro ? Jan 24 10:35:39 and include familiar ? Jan 24 10:35:53 evening Jan 24 10:36:15 hey Jan 24 10:36:21 hi reenoo Jan 24 10:36:44 hi zecke Jan 24 10:36:46 hey pb_ Jan 24 10:40:04 Ifaistos: There's no one "right" way of doing that Jan 24 10:45:09 ok. Jan 24 10:45:35 are all the parameters used in the conf files documented somewhere ? Jan 24 10:58:41 Ifaistos: if you want to customize an image, I recomment you to take gpe-image or opie-image and modify them (in a diferent directory, don't break the originals) Jan 24 10:59:16 for example, to have a customized version of gpe-image, chaging packages in meta-packages.bb could be a start Jan 24 10:59:32 I'm a noob, if I'm wrong someone corrects me plz Jan 24 11:12:59 god damn retarded udev Jan 24 11:19:44 anyone has the link to SL6000 ROM backup instructions? Jan 24 11:46:42 RP: got basic NIC configuration via /etc/network/interfaces working Jan 24 11:48:04 I just assume that we do not need the bootp stuff.. Jan 24 11:49:04 CoreDump|home: Anything is better than nothing :) Jan 24 11:50:04 btw: "reboot" does a "halt" here heh Jan 24 11:50:04 hi Jan 24 11:50:26 hi Skyhusker Jan 24 11:50:42 hey CoreDump|home Jan 24 11:50:46 CoreDump|home: I wish people would stop mentioning that... :) Jan 24 11:50:52 lol Jan 24 11:52:21 RP: have you seen this patch that speeds up pcmcia? I get about a 30% speed increase. http://www.home.arm.linux.org.uk/developer/patches/viewpatch.php?id=2824/1 Jan 24 11:52:33 doesn't look like it was discarded for any good reason Jan 24 11:53:27 btw, the cvs uri for quilt is wrong Jan 24 11:53:34 or at least, doesn't work for me right now Jan 24 11:53:59 the solution is s/cvsroot/sources, according to http://savannah.gnu.org/forum/forum.php?forum_id=4168 Jan 24 11:54:42 RP: on my CF card I go from about 2.12MB/s to 2.77MB/s on pxa255 Jan 24 11:55:55 mreimer: I hadn't seen that, it looks interesting. The disgarded reasons are just rmk being his usualy self (and the spacing is valid critisim). Have a look at the pile I've written that have been disgarded (and almost every one later accepted) ;-) Jan 24 11:56:06 :-) Jan 24 11:56:24 free speed Jan 24 11:56:55 Not often that happens :) Jan 24 11:59:54 what the hell is wrong w/ vi in OZ Jan 24 12:01:07 someone please do a "which cardcl" on OZ for me Jan 24 12:02:25 CoreDump|home: cardcl? Jan 24 12:02:33 t Jan 24 12:02:38 ctl Jan 24 12:02:52 Its in /sbin Jan 24 12:03:03 ty Jan 24 12:03:48 heh. git depends on rcs? Jan 24 12:11:13 hi all Jan 24 12:11:22 hey koen Jan 24 12:11:29 hey koen|away Jan 24 12:13:33 any news in OE land? Jan 24 12:14:13 hey koen Jan 24 13:09:37 I want to make a .bb for software that only needs "./configure; make". How to do that? Can you point to a simple example? Jan 24 13:10:13 inherit autotools Jan 24 13:10:51 wow, that's easy. thanks reenoo Jan 24 13:12:12 np Jan 24 13:12:52 it does a bit more than just ./configure though. it runs autoreconf to regenerate the configure script (against our own macros) Jan 24 13:18:01 ok Jan 24 14:04:12 wb kergoth`work :) Jan 24 14:38:42 where's the proper place to post new .bb files? Jan 24 14:39:51 bugzilla Jan 24 14:40:10 thanks again reenoo Jan 24 14:40:48 I don't know about the current dns situation though. bugs.oe.org may not resolve ATM. Jan 24 15:00:00 hi all Jan 24 15:00:16 hi florian Jan 24 15:00:17 kergoth`work: ping Jan 24 15:01:44 florian: pong Jan 24 15:01:58 hi kergoth Jan 24 15:02:03 hey Jan 24 15:04:50 kergoth`work: hi, any idea why openembedded.org doesn't resolve anymore? Jan 24 15:05:24 not offhand, i'll take a look when i get back to the hotel. finally have net access there Jan 24 15:05:48 (its wireless only, and i've got no wireless card for the desktop, and no laptop, so had to borrow a usb wifi adapter from a guy at work.. heh) Jan 24 15:06:00 kergoth`work: do you have an opinion on the big OE SCM shoot-out/bake-off? Jan 24 15:06:40 havent been on the lists or anything, no idea whats going on there Jan 24 15:07:15 kergoth`work: not much. I use tailor to create git,darcs,svn,mercurial,bazaar trees Jan 24 15:07:39 kergoth`work: and crofton (svn) and RP (git) set up servers and people can test the SCM with real data Jan 24 15:07:57 and maybe they think it is better than monotone and in that case we might switch... Jan 24 15:08:00 ahh nice, so they fixed some of the tailor monotone issues i take it Jan 24 15:08:03 cool Jan 24 15:08:21 i'll take a look when i get back to the hotel. i havent spent much time with git or mercurial, need to try those Jan 24 15:08:32 kergoth`work: well if we switch (find no compelling reason to not switch) I will hack on tailor to be a bit better Jan 24 15:08:58 kergoth`work: alwin is preparing a server where I ca either host darcs or mercurial or both :} Jan 24 15:09:01 I see you are still pushing stuff into my svn server :) Jan 24 15:09:22 kergoth`work: and I'm using svk+svn to merge to croftons server Jan 24 15:09:23 i'm not a big fan of bazaar or darcs, neither really seemed up to par when i used them last, but i've been out of it for a while Jan 24 15:09:27 i /love/ svk :) Jan 24 15:09:38 kergoth`work: so svk -I takes ages... Jan 24 15:09:46 svk smerge -I -B that is Jan 24 15:10:07 zecke, that means push one svn repo to another with rev info? Jan 24 15:10:56 Crofton: yes, apply each rev separately... I should have given you a tarball with the revs and you should have issued the commands Jan 24 15:11:06 heh Jan 24 15:11:20 hindsight is always good Jan 24 15:11:37 now we know not to populate servers this way :) Jan 24 15:12:18 kergoth`work: darcs has issues due the usage of ghc... Jan 24 15:12:26 kergoth`work: git+mercurial require ssh logins Jan 24 15:12:53 http://oe.handhelds.org/cgi-bin/moin.cgi/ScmVotingProcedures#preview Jan 24 15:13:08 http://oe.handhelds.org/cgi-bin/moin.cgi/SvnScmTrial?action=show Jan 24 15:13:25 http://oe.handhelds.org/cgi-bin/moin.cgi/GitScmTrial?action=show Jan 24 15:13:36 everything is still to be discussed, specially the selection phase Jan 24 15:13:41 I'm off to bed! Jan 24 15:13:46 kergoth`work: have fun! and take care! Jan 24 15:14:02 gn Jan 24 15:14:10 'night zecke Jan 24 15:16:01 ~lart braindamaged assignments Jan 24 15:16:02 * ibot cats /dev/urandom into braindamaged assignments's ear Jan 24 15:16:28 reenoo: do you know if a PriorityQueue can be resorted? Jan 24 15:16:48 depends on the underlying data structure Jan 24 15:16:59 reenoo: the Java one Jan 24 15:17:19 java.util.Prio (with a custom Comperator set) Jan 24 15:17:45 ah, haven't used that yet. probably not. Jan 24 15:18:25 could remove elements from one instance and add them to another Jan 24 15:18:40 should be O(n log n) Jan 24 15:18:54 unless their implementation sucks :) Jan 24 15:18:59 ~lart sun Jan 24 15:19:00 * ibot strangles sun with a 9-pole serial cable Jan 24 15:19:22 reenoo: I have fun implementing the dijkstra ( I do not want to hack it in 5 lines of codes...) Jan 24 15:19:41 good nite Jan 24 15:19:54 dijkstra's shortest path algorithm? Jan 24 15:20:00 yeah :} Jan 24 15:20:28 ah, IIRC the trick was to ignore duplicates in the queue or something Jan 24 15:20:45 not exactly clean but it works >:) Jan 24 15:21:07 ah well, keys can only decrease in dijkstra anyway Jan 24 15:21:42 right Jan 24 15:22:14 some tutors started bitching about that though... Jan 24 15:23:00 well we embed that into our E-Chalk for Demonstration Jan 24 15:23:11 so you draw some circles, connect them with edges Jan 24 15:23:20 press the start button and see dijkstra working Jan 24 15:23:36 sounds cute Jan 24 15:24:13 reenoo: our bubble sort is even more cute. You draw lines on the board Jan 24 15:24:22 reenoo: hit play, and you see them moving around... Jan 24 15:24:43 i really need to get around to reading the art of programming. dont know very many of the common algos Jan 24 15:25:15 well I have the Cormen Jan 24 15:25:30 cya Jan 24 15:26:16 taocp is good for recommendations. use algorithm foo to solve problem bar. Jan 24 15:26:55 in this case use a leftist tree for the priority queue Jan 24 15:28:02 hm maybee zecke has my cormen and he didnt told me Jan 24 15:48:19 oh well. it's just not right when you have to check what the reference implementation (binaries only, no source code) does to understand what you're supposed to do :/ Jan 24 16:58:57 Is there a way to access oe bugzilla without openembedded.org? Jan 24 17:02:06 You might find it works if you add a local /etc/hosts entry for bugs.openembedded.org pointing to 192.216.230.225 Jan 24 17:02:46 okay :-) Jan 24 17:14:30 if that fails, I guess you have to send hate-mail to Greg. Jan 24 17:15:06 :-) Jan 24 17:15:16 it worked fine... Jan 24 17:15:27 rock Jan 24 17:15:30 well... in fact it still works. Jan 24 17:17:59 okay, very good Jan 24 17:19:51 * pb__ go to sleep now Jan 24 17:20:16 'night pb__ Jan 24 17:20:23 good call actually Jan 24 17:20:27 'night all Jan 24 17:20:34 need to be at the office for some ghastly meeting at an early hour in the morning Jan 24 17:20:39 night all Jan 24 17:22:49 same here... Jan 24 17:22:52 good night Jan 24 21:02:54 re Jan 24 21:36:03 kergoth: what do you see as the key features of svk that set it apart from the pack? Jan 24 21:38:07 hmm, let me think. its been a while since i even used it, what with the move and all, and even longer since i used the alternatives. it has star merges and cherry picking, but afaik those are supported by most of the options. i'd say one of the main advantages is that its based on svn. you can get advanced features without modifying the upstream svn repository, and svn is known to be solid and stable. it can also mirror upstream cvs Jan 24 21:38:13 i dunno, ask me again Jan 24 21:38:19 after i've had a chance to poke at the tailor repositories zecke did Jan 24 21:38:20 :) Jan 24 21:39:04 kergoth: thx Jan 24 21:40:33 kergoth: I'm looking for a CM solution for a new company, where most of the employees (doing software tools and modelling of SoCs) have come from a ClearCase background. I'm looking to see whether there is an open source solution which will meet the needs. Jan 24 21:41:30 The needs seem to line up well with OE and nslu2-linux requirements for a distributed CM system, so I'm keenly interested in this bake-off. Jan 24 21:42:12 i'll be interested as well, the folks at my new job are using clearcase and pondering possible alternatives Jan 24 21:44:34 i /love/ clearcase's config specs Jan 24 21:44:34 they dont really line up well with the idea of a changeset though. they tend to be file oriented Jan 24 21:55:55 yeah, clearcase doesn't really have the idea of a changeset, apart from the derived object and configuration record stuff. Jan 24 21:56:10 (which you get through audited builds with clearmake) Jan 24 21:57:54 there's an addition to clearcase called UCM that adds "activities" which are a changeset, and adds development streams and the like, but its mediocre. Jan 24 21:58:09 :\ Jan 24 21:58:42 anyone here use vim in windows (not cygwin)? need to know where it expects the user's vimrc and such. heh Jan 24 22:02:34 ah. ~/_vimrc and ~/vimfiles/ Jan 25 00:25:41 goodmorning to all ! Jan 25 01:58:50 morning all Jan 25 01:59:22 hi ! Jan 25 02:00:04 RP: Sorry to bug you so early but i hit (another) snag... Jan 25 02:00:42 RP : I am trying to build libxml-libxml-common-0.13 Jan 25 02:00:56 and fails during the config Jan 25 02:01:17 ERROR: function do_configure failed Jan 25 02:01:54 the configure is using an xml2conf which is supposed to return the path to libxml2 Jan 25 02:02:04 libxml is compiled Jan 25 02:02:16 but xml2conf says it can not find it Jan 25 02:02:52 sorry not the path... if libxml2 exists Jan 25 02:04:41 should i use an EXTRA_OECONF defining the path to the libs ? Jan 25 02:12:06 morning Jan 25 02:16:12 hi XorA Jan 25 02:16:38 Ifaistos: EXTRA_OECONF might help, yes Jan 25 02:16:57 Ifaistos: Some packages are more tested that others :-/ Jan 25 02:17:30 hi all Jan 25 02:18:15 libxml is installed here oe/feux/build/tmp/staging/i586-oe-linux/lib/ Jan 25 02:18:59 but i am confused on how the dir should be declared in EXTRA_OECONF Jan 25 02:20:53 Ifaistos: ${STAGING_LIBDIR}might help Jan 25 02:35:00 morning all Jan 25 02:47:10 morning **** ENDING LOGGING AT Wed Jan 25 02:59:57 2006