**** BEGIN LOGGING AT Thu Sep 16 02:59:57 2010 Sep 16 03:07:03 down git server ? I can not pull . Sep 16 03:07:50 Hello, im getting: git.openembedded.org[0: 140.211.169.165]: errno=Connection refused Sep 16 03:08:03 also: fatal: http://git.openembedded.org/info/refs not found: did you run git update-server-info on the server? Sep 16 03:09:33 is it normal? am I doing something wrong or just a temporary problem with the server? Sep 16 03:15:08 marcompile: I have same problem now. Sep 16 03:35:23 hmm... "How to submit a patch for dummies" probably needs a different name. If you refer someone there you might end up insulting them :) Sep 16 03:48:58 grg: heh thought came to me Sep 16 03:49:37 iwamatsu: git should be up now Sep 16 04:04:13 grg: I renamed the page Sep 16 04:04:26 khem, :) Sep 16 04:04:44 http://wiki.openembedded.org/index.php/How_to_submit_a_patch_to_OpenEmbedded Sep 16 04:05:53 How_to_submit_a_patch_to_OpenEmbedded_for_those_with_manual_allergies ? Sep 16 04:47:37 hmm . . . anyone else having trouble pulling from the oe repo? Sep 16 04:47:56 I get this: http://pastebin.com/NKbpnZLh Sep 16 04:47:56 sakoman: it should be fixed Sep 16 04:48:32 khem: still broken for me Sep 16 04:51:02 sakoman: works ok for me just now Sep 16 04:51:04 me too. Sep 16 04:51:10 < broken Sep 16 04:51:41 sakoman: can you try a clone Sep 16 04:52:00 khem: ok, will try a clone Sep 16 04:52:15 I checked by clone command, but can not clone. Sep 16 04:53:33 khem: same error Sep 16 04:53:52 khem: fatal: unable to connect a socket (Connection refused) Sep 16 04:53:59 me too. Sep 16 04:57:14 time for sleep now, will try again in the morning Sep 16 04:57:25 sakoman: strange works ok for me Sep 16 04:58:07 khem: it work for me earlier today, but has been broken since I tried an hour or so ago Sep 16 04:58:30 sakoman: there was mail about system upgrade Sep 16 04:58:46 but upgrade was done successfully Sep 16 04:58:54 so it should have been back to normal Sep 16 05:00:53 well, still broken here. hopefully tomorrow morning it will be all better :-) Sep 16 05:01:49 hmmm Sep 16 05:02:01 browsing cgit.openembedded.org works fine, but no pulls or clones work Sep 16 05:02:20 did the server ip address change? Sep 16 05:02:51 if so, perhaps the change hasn't propagated to me yet Sep 16 05:04:16 140.211.169.165 Sep 16 05:04:31 is what it shows for me Sep 16 05:04:32 yeah, that's what I get too Sep 16 05:04:40 so that shoots that theory Sep 16 05:05:14 perhaps it is just my ip addresses that are blocked :-) Sep 16 05:06:23 iwamatsu: does your "hmmm" mean that you have learned something new? Sep 16 05:06:44 I changed address of repo to git://repo.or.cz/openembedded.git. work fine. Sep 16 05:07:07 sakaman: no. Sep 16 05:07:27 hmmm fresh clone same issue for me Sep 16 05:56:42 i get this with loads of X apps under kdrive: "Warning: Unable to load any usable fontset" anyone know how to fix this? Any particular font recipe i should be including in my images? Sep 16 05:58:11 may be fontpaths are not set Sep 16 05:58:13 xset fp rehash Sep 16 06:06:16 khem, thanks, but no cigar Sep 16 06:16:03 tharvey: warflyr: backreading the log, and wrt upnp: if you want media serving etc have a look at coherence (recipe python-coherence) Sep 16 06:18:36 btw gm everyone Sep 16 06:20:22 morning eFfeM_work Sep 16 06:20:51 grg you missed a zziplib user, see ML Sep 16 06:22:12 the list is a bit sluggish right now... Sep 16 06:22:27 ogre/ogre-egl_svn.bb:DEPENDS = "gtk+ wgois cppunit zziplib boost freeimage freetype virtual/libx11 virtual/egl" Sep 16 06:23:15 hmmm... not sure how i missed that. Should be ok with the newer zziplib recipe Sep 16 06:25:20 i'll find out tomorrow when bitbake -k world progresses again Sep 16 06:26:05 it will be really nice to use bitbake world for finding build failure regressions. Sep 16 06:26:06 grg, about the xtscal crash and xcb mismatch thing, i don't know enough how to debug which libs are in conflict. i'll probably just give up and take a nap :) Sep 16 06:26:28 htns, naps are awesome. I fully support that option. Sep 16 06:41:43 does anybody know if the beagleboard-xM has the TI HECC CAN-BUS controller? Sep 16 06:42:13 good morning Sep 16 06:42:18 hi Sep 16 07:08:04 grg, bitbake world for regression would be fine will start a new test tomorrow, unfortunately the server I started this on was new and still had some issues, so it is being reinstalled Sep 16 07:08:26 still have some issues with virtual/egl etc though Sep 16 07:09:11 foobar___: this is hardly an appropriate channel for that q, better try #beagle or read the doc Sep 16 07:20:57 hi cyberdeck Sep 16 07:27:29 morning Sep 16 07:29:23 hi hrw Sep 16 08:08:28 hi blinder Sep 16 08:40:41 Q: is it possible to use two download dirs? I would like to have one as central r/o and one with local stuff Sep 16 08:45:24 unionfs :) Sep 16 08:46:00 mount central r/o dir from storage and your local dir above :) Sep 16 08:49:45 Jay7: yes, didn't consider that, actually was hoping there was a way to specify two dirs or so Sep 16 08:50:34 eFfeM_work: I'm sure there is no ready way Sep 16 08:51:53 you can try to use local mirror Sep 16 08:52:13 but all new downloads will go to your local directory anyway Sep 16 08:53:07 Jay7: I already have a local mirror to speed things up but the issue is local dirs being populated with it again Sep 16 08:53:39 eFfeM_work: ask some bitbake guru then :) Sep 16 08:54:04 may be it easy enough to implement Sep 16 08:54:14 that's why I popped the Q here, whether it already existed Sep 16 08:56:43 hi woglinde Sep 16 09:34:06 hi Sep 16 09:34:54 hi ericben Sep 16 09:35:58 hrw: seems from removal.txt you proposed to remove Maemo on 2006-12-15, we still have recipes/maemo (but half of maemo seems to be in nonworking) (and we also have recipes/maemo4) Sep 16 09:36:03 is this still of any use ? Sep 16 09:36:06 hi ericben Sep 16 09:36:44 eFfeM_work: some maemo components were updated, maemo5 ones should even build Sep 16 09:36:53 eFfeM_work: drop that part from removal.txt Sep 16 09:37:01 ok, will do Sep 16 09:37:36 hi woglinde eFfeM_work Sep 16 09:38:10 hrw, there you go Sep 16 10:14:38 hm, cia does not seem to be workign Sep 16 10:16:11 anyone has an idea on how to add an entry into some rc file, so that i may run an arbitrary script on system startup. i would like to do this in a recipe function, such as pkg_postinst() Sep 16 10:45:35 git server seems down for clone Sep 16 10:57:41 ericben: i always thought git servers had to be up for cloning ;) Sep 16 10:57:44 scnr Sep 16 10:57:53 Is git.openembedded.org down? Sep 16 11:01:59 Is git.openembedded.org down? Or is it my firewall? Sep 16 11:03:42 no idea... maybe try the browser to access it Sep 16 11:06:40 i can not manage to clone the tree from the browser interface. Is it possible? Sep 16 11:09:35 hey. from the guide here i see that the default kernel for beagleboard should be 2.6.29, but on the image i've done with OE i get 2.6.32, although the 2.6.29 recipe has DEFAULT_PREFERENCE_beagleboard="1" and my MACHINE="beagleboard" Sep 16 11:09:47 why did OE build 2.6.32 ? Sep 16 11:10:12 for me it worked an hour or two ago Sep 16 11:10:38 CMoH: because your distro pins it to .32 Sep 16 11:10:43 cmoh default for beagleboad os 2.6.32-psp Sep 16 11:10:50 and never believe guides, they are in 99% of the time outdated Sep 16 11:11:23 thanks; well, if i started from scratch asking here you'd be pointing fingers at me for asking not "smart-questions" :D Sep 16 11:12:02 cmoh without psp-kernel yout dont have dsp und sgx support Sep 16 11:12:35 Is everybody accessing git.openembedded.org normally? Sep 16 11:12:41 I can not... Sep 16 11:12:51 thanks again Sep 16 11:13:07 ozgur yeah strange problem Sep 16 11:13:12 here it worked too Sep 16 11:13:20 maybee we have to wait for ka6sox Sep 16 11:15:02 thanks Sep 16 11:15:22 another q: what is the difference between linux-omap-psp and linux-omap-multi-psp Sep 16 11:15:23 ? Sep 16 11:21:02 CMoH: that is probably better asked in #beagle Sep 16 11:21:17 ok, thanks Sep 16 11:31:52 CMoH multi made different kernels with one recipes Sep 16 11:31:59 if I remember correctly Sep 16 11:32:29 you mean with different patch sets? Sep 16 11:32:39 diffrent config Sep 16 11:32:44 cpuidle enabled Sep 16 11:32:59 an one with musb/g_ether build in Sep 16 11:33:07 i see Sep 16 11:33:14 so the kernel dont segfault when powerup via usb Sep 16 11:38:15 florian, ping Sep 16 11:38:37 Crofton: pong Sep 16 11:38:42 hey Sep 16 11:38:52 I am going to send a really quck G announcement Sep 16 11:39:01 otherwise I will never get it done Sep 16 11:39:02 :) Sep 16 11:39:07 hm I will not made it to oedem this year Sep 16 11:39:11 There are two agenda items I can think of Sep 16 11:39:14 yes i think its fine Sep 16 11:39:20 Financial report and board election Sep 16 11:39:25 right Sep 16 11:39:36 for the GA is there anything else we HAVE to do Sep 16 11:39:45 most other stuff can be done as OEDEM Sep 16 11:39:47 we can add more later, it doesn't have to be a fixed agenda Sep 16 11:39:52 yeah Sep 16 11:39:55 right Sep 16 11:42:05 and voting in new members :) Sep 16 11:52:26 libgpod build failing: http://pastebin.com/nHMSLwcX Sep 16 11:52:36 it worked last week Sep 16 11:54:50 hmmm Sep 16 11:54:59 depends? Sep 16 11:55:50 anyone know what provides the gdk-pixbuf.h header? Sep 16 11:59:00 ah gdk-pixbuf :) Sep 16 12:24:13 more people need to do clean builds on multi core machines ..... Sep 16 12:35:23 I often do Sep 16 12:51:36 hi all Sep 16 12:54:46 gm Sep 16 12:56:52 can anyone tell me how i can view a patch (with status != new) in the patchwork Sep 16 12:57:53 oh, just figured out Sep 16 13:47:45 my git config likes to update master Sep 16 13:47:47 annoying Sep 16 13:51:35 Is something not well at git.openembedded.org? I can ping, but not fetch. Sep 16 13:54:23 tasslehoff: git pull works here Sep 16 13:59:14 eFfeM_work: hm. connection refused here. Sep 16 13:59:32 others have mentioned this too, no idea why Sep 16 13:59:38 are you pulling ro or rw Sep 16 14:00:39 eFfeM_work: I think ro. Sep 16 14:00:59 might be related to that, the other personthat mentioned thsi was also ro Sep 16 14:01:10 try ro, or if you are in eu use repo.or.cz Sep 16 14:04:21 eFfeM_work: we're outside my git-knowledge now :) Sep 16 14:04:36 tasslehoff: do you have commit right s? Sep 16 14:04:50 if not you are definitely pulling ro Sep 16 14:05:09 can I just change upstream to be repo.or.cz? Sep 16 14:05:17 no commit rights Sep 16 14:05:21 i think so in your .git/config file Sep 16 14:05:49 i have this Sep 16 14:05:51 [remote "origin"] Sep 16 14:05:51 fetch = +refs/heads/*:refs/remotes/origin/* Sep 16 14:05:51 url = git://repo.or.cz/openembedded Sep 16 14:07:17 eFfeM_work: that did the trick. thanks Sep 16 14:07:23 yw Sep 16 14:29:07 hrm, what's the proven way today to put /var on a tmpfs with all its directories? Sep 16 14:29:13 pb_: hey Sep 16 14:29:23 hi zecke Sep 16 14:29:26 hi blindman Sep 16 14:29:54 pb_: thanks for the botmail. I was trying to setup a backup MX, I thought picking the same prio for the MX record is fine :) Sep 16 14:30:15 pb_: and I should have used a 4xx instead of the 5xx for the user. i think the bouncing is fine now Sep 16 14:30:27 zecke: ah, very good Sep 16 14:34:30 gm all Sep 16 14:35:28 hi khem Sep 16 14:35:29 some openmoko recipes are broken e.g. openmoko-libs as the sources upstream are organised differently than when the recipe must have been wwritten Sep 16 14:35:33 hi ericben Sep 16 14:35:54 whats should we do with them do we still need to maintain them Sep 16 14:37:07 pb__: thanks for noticing me, could you resend the (personal) mail you were sending? Sep 16 14:49:28 pb__: do you know if someone volunteered to write notes for the 2nd meeting 2nd attempt? Sep 16 14:49:40 zecke: will do. let me see if I can find it now. Sep 16 14:49:57 zecke: not to my knowledge. Sep 16 14:52:15 pb__: do you happen to have the chat log of the 2nd meeting? Sep 16 14:52:20 zecke: yup Sep 16 14:52:23 do you want me to mail it to you? Sep 16 14:52:55 pb__: yes, please... then I can write notes tonight Sep 16 14:54:27 zecke: done Sep 16 14:56:50 zecke: I also forwarded what I think was the email that bounced. Sep 16 14:57:02 not 100% sure though because, stupidly, I didn't keep the original UDN Sep 16 14:57:16 pb__: thanks a lot Sep 16 15:01:55 you guys aren't very technical ... Sep 16 15:10:03 anyone seeing gdk-pixbuf weirdness? Sep 16 15:10:26 such as? Sep 16 15:11:23 I'll post after lunch Sep 16 15:11:29 wiped tmp and started over Sep 16 15:11:34 I had done some hacking aobut Sep 16 15:39:44 * fahad wondering if there's a difference in checking out a repo from git://git.openembedded.org/openembedded or from http://repo.or.cz/r/openembedded.git Sep 16 15:39:58 eFfeM_work, thx for the note on python-coherence - that does seem very full-featured and does have 'Puncher' to manipulate IGD's. I think I'll still try to post a recipe for miniupnp however for those highly embedded systems that may not have room for python Sep 16 15:42:45 khem ping Sep 16 15:44:04 is git down? Sep 16 15:44:53 I know that CIA is broken but git shouldnt' be Sep 16 15:55:32 who admins the git server again? Sep 16 16:00:20 RP: hmm, thinking BuildStarted should fire after runqueue generation, not before, as that's more true to when the build is actually beginning.. runqueue generation is a preparation step, not a part of the actual building. thoughts? Sep 16 16:06:36 kergoth_, you mean gitosis or the underlying infra? Sep 16 16:07:11 underlying. i have gitosis admin, but i want to look into why bitbake cia notifications don't work, which requires shell access Sep 16 16:07:15 unless someone else wants to fix it :) Sep 16 16:07:37 bitbake and oe, for that matter, though the oe one might be on the CIA configuration end of things Sep 16 16:07:41 I do and am working to fix hte link that is broken Sep 16 16:07:43 will check that Sep 16 16:07:53 okay, well bitbake cia notifications have never worked, i don't think Sep 16 16:08:01 its hooks must be set up differently than oe's Sep 16 16:08:05 or its on the cia.vc end.. Sep 16 16:08:07 but i don't think so Sep 16 16:08:13 oh...didn't know that. Sep 16 16:08:20 cia hasn't received any updates for bitbake in ages, if you look at the website Sep 16 16:08:35 once I figure out how to fix the oe ones I can look into adding one for BB Sep 16 16:08:38 no rush, but if you have a chance, would appreciate your looking into it Sep 16 16:08:40 cool Sep 16 16:08:41 thanks Sep 16 16:08:45 np Sep 16 16:10:01 okay, yeah, verfiied it must be on that end for oe, if you look at http://cia.vc/stats/project/openembedded, it has no commit info for other branches, so its not just it receiving it and not showing it in the channel (which is a possibility, depending on the configuration of the filters for the channel) Sep 16 16:10:06 s/that end/the server end/ Sep 16 16:10:52 there is a broken link to /usr/bin/mail Sep 16 16:10:53 in /etc/alternatives Sep 16 16:11:01 ah Sep 16 16:11:07 right now I'm doing db recoveries and then that will be fixed Sep 16 16:11:45 not sure if you know, but there are cia notifier scripts that notify via xml/rpc rather than email -- can be more reliable, if the xml/rpc modules are available for whatever language the script is written in Sep 16 16:11:48 cool Sep 16 16:15:00 hi kergoth Sep 16 16:15:29 hey Sep 16 16:15:35 some of the patches that i sent in the last couple of weeks are still in the patch queue Sep 16 16:15:55 that's pretty common, it can take a while. i have some that have been in queue for months :\ Sep 16 16:16:12 hmm... Sep 16 16:16:16 i don't like acking commits from coworkers, not necessarily unbiased, prefer letting others do it Sep 16 16:16:27 is there a way to bump them up? Sep 16 16:16:48 can reply to the mails, so the threads show back up in peoples inboxes Sep 16 16:16:51 might be enough Sep 16 16:17:10 right, Sep 16 16:17:34 with recipes, its easier to get reviewers, as most people are comfortable enough with them -- if you touch a class, god help you, no one is confident enough to risk breaking things :) Sep 16 16:17:37 * kergoth_ mutters Sep 16 16:21:21 khem, what was the script that you told me about line 154 in? Sep 16 16:52:05 remote: Cannot exec /usr/bin/mail at /home/git/git-notify line 154. Sep 16 16:52:12 ka6sox: that was the one Sep 16 16:59:08 * khem has meeting Sep 16 17:02:40 good morning/day/evening/night everyone :-) Sep 16 17:03:59 why does it feel like a bunch of stuff that depends on python-native is broken atm Sep 16 17:05:02 http://pastebin.com/PRJCzuhK Sep 16 17:06:03 because (at least for angstrom) python was changed from 5.10.1 Sep 16 17:06:10 from 5.8.8 to 5.10.1 Sep 16 17:06:48 oops, taht was python, not perl Sep 16 17:06:55 sorry for the confusion Sep 16 17:07:19 anyway, koen did this for angstrom, http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=8662dc3a3f0c952659501ac16a48a0108179c331 Sep 16 17:07:35 not sure if that will not break the libperl recipes Sep 16 17:08:15 i didn't have python issues (did a clean build a few days ago) Sep 16 17:10:01 || amke and multuiple BBTHREADS? Sep 16 17:10:48 PARALLEL_MAKE="-j4" Sep 16 17:10:48 BB_NUMBER_THREADS = "2" Sep 16 17:11:00 but note i have build minimal, not angstrom Sep 16 17:11:29 this version Sep 16 17:11:30 ../tmp_minimal/work/i686-linux/python-native-2.6.5-ml11.0: Sep 16 17:16:59 Crofton using angstrom : NOTE: package python-docutils-native-0.5-ml1: task do_build: Succeeded Sep 16 17:22:37 I have || amke and BBTHREAD in place Sep 16 17:22:43 and am using 2010.x Sep 16 17:22:54 I suspect it is just things missing DEPENDS Sep 16 17:23:21 possibly if recenttly converted to BBCLASSEXTEND Sep 16 17:23:51 but that happened a while ago in this case Sep 16 17:24:18 it is pretty clear that python needs to be added to things that need it Sep 16 18:01:12 Crofton : || make = 8 BBTHREAD = 4 angstrom 2008.1 here Sep 16 18:02:58 if my build works, I'll push the DEPEND update Sep 16 18:03:18 I suspect there are a number of packages that have bad depends, that work just by luck Sep 16 19:03:04 re Sep 16 19:46:26 Crofton|work: I think if someone adds a new recipe then he should do bitbake from scratch and see if it builds Sep 16 19:47:10 if we enforce that then in future recipe will become more free of false deps or missing deps Sep 16 19:49:08 yeah Sep 16 19:49:14 the problem is always time Sep 16 19:49:25 and changes to recipes can do the same thing Sep 16 19:49:38 Crofton|work: you do it now or later Sep 16 19:49:49 Crofton|work: amount of time is always same Sep 16 19:50:09 but grief it can cause varies Sep 16 19:50:15 yeah Sep 16 19:50:34 I always think doing a good job upfront saves a lot of grief Sep 16 19:50:59 upstream does not need a new change right away so it can be tested Sep 16 19:51:38 ah this laptop hd is too loud I think its going to die soon Sep 16 19:52:00 I have a problem wiht a recipe that creates installation dir in the wrong place from which i get error stating it cant find a certain file Sep 16 19:52:15 time for a Fry's trip Sep 16 19:52:17 TheGreat88: is that a question or a riddle Sep 16 19:52:28 lol to me riddle Sep 16 19:52:31 Crofton|work: yeah Sep 16 19:52:40 let me try n explain Sep 16 19:53:01 i am trying to bake a dsplink module for the over board Sep 16 19:53:56 Crofton|work: the clone of this hd just finished Sep 16 19:53:58 there are 3 file from ti xdctools, cgtools and bios that have a path file which is suppose to set the installation dir Sep 16 19:54:03 so I can happily remove it Sep 16 19:54:47 but it creates the dir in the wrong place Sep 16 19:55:08 TheGreat88: where is dir suppose to be and where does it create Sep 16 19:55:18 tried the workaround--manually putting the .bin file where its suppose to go Sep 16 19:55:27 it puts it one before Sep 16 19:56:14 | cp: cannot stat `/home//oe/tmp/work/overo-angstrom-linux-gnueabi/ti-xdctools-3_16_01_27-r2/xdctools_3_16_01_27 Sep 16 19:56:20 this is empty Sep 16 19:56:35 the file is in | cp: cannot stat `/home/amir/oe/tmp/work/overo-angstrom-linux-gnueabi/ti-xdctools-3_16_01_27-r2/xdctools_3_16_01_27 Sep 16 19:57:19 made a mistake of putting it there manually so it did compile but then do_preparesources fails Sep 16 19:58:29 this might be the reason and patch here but im not sure: http://permalink.gmane.org/gmane.comp.handhelds.openembedded/36135 Sep 16 19:59:19 TheGreat88: test this patch and provide info to ml Sep 16 19:59:23 if it works for you Sep 16 19:59:27 we can apply it Sep 16 19:59:35 cool Sep 16 19:59:37 will do Sep 16 19:59:41 thanks Sep 16 20:00:20 khem, actually you should not only build the recipe from scratch, you should build it from scratch with pstage outside tmp, then rm tmp then bake again Sep 16 20:00:55 one of the most problematic issues is that some depends are not made explicit because e.g. some toolchain recipe already drags it in Sep 16 20:01:25 once you delete tmp it gets populated on demand Sep 16 20:01:41 so its effect is same with or w/o pstage Sep 16 20:03:47 is it? what about recipes that are intermediates? e.g. does gcc initial gets populated if gcc is already in pstage ? Sep 16 20:04:01 if you understand what I mean Sep 16 20:07:47 eFfeM: intermediate is for bootstrapping toolchains Sep 16 20:07:51 its special case Sep 16 20:08:06 generally packages added are applications Sep 16 20:08:37 khem, maybe I explained wrong, let me give another example Sep 16 20:08:53 eFfeM: for toolchain there should be a work around may be we should only have SDKs and use SDKs for internal toolchain too Sep 16 20:09:15 A depends on B which depends on C Sep 16 20:09:53 fine whats issue with that Sep 16 20:09:54 if tmp is removed with pstage outside tmp and A is rebuild will B only be retrieved from pstage or will C also be retrieved ? Sep 16 20:10:06 can't type that fast :-) Sep 16 20:10:29 I think a should drag b and b should drag c Sep 16 20:10:38 this seems new: Sep 16 20:10:41 NOTE: multiple providers are available for runtime libstdc++-dev (gcc, external-toolchain-csl, external-toolchain-generic, external-toolchain) Sep 16 20:11:20 khem i kinda assumed only those packages that are in the direct depends should be restored, but thinking of it the direct depends at least would need its rdepends Sep 16 20:11:35 Crofton|work: when was your last clean rebuild ? Sep 16 20:11:44 an hour ago Sep 16 20:11:55 Crofton|work: one of these packages is providing something thats different Sep 16 20:11:57 well a bit longert, but you get the idea Sep 16 20:12:49 ah ok that outrules that you still have the old stdc++ patch (and I assume you use dev head) Sep 16 20:13:27 actually i also had an issue that someone dragged in libc6-dev, no idea which package haven't found the cause yet Sep 16 20:13:28 one of these days I should look into separating gcc runtime from gcc itself Sep 16 20:13:36 that will solve these problems Sep 16 20:13:43 would be nice Sep 16 20:14:00 poky has it Sep 16 20:14:03 yeah Sep 16 20:14:04 it's just not elligant :) Sep 16 20:14:22 Tartarus: I think with OE it never will be Sep 16 20:14:51 unless we have a way to package cross and target ipks out of cross recipes Sep 16 20:15:06 its hard to separate libgcc from gcc Sep 16 20:24:44 tinderbox down? I get: NOTE: oestats: error starting build, disabling stats Sep 16 20:29:09 hmm, I wonder if I ahve a hack that includes libsdc++ Sep 16 20:29:24 I am close to dev, but have some of my baggage involved Sep 16 20:31:41 Crofton|work: well what I was thinking about was this patch: http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=d888b42cc4597814b714a8c008df7094c7de53d1 Sep 16 20:32:05 if you do not have that one, but you do have the other fix before it then this might be the case Sep 16 20:32:38 or if your build is older than the one from the above commit as that (in my opinion erroneously) did not have a PR bump Sep 16 20:32:55 clean build Sep 16 20:33:14 yes you said so Sep 16 20:33:25 but you also said close-to-dev, dunno how close that is Sep 16 20:33:33 so thought i'd mention it anyway Sep 16 20:33:34 today Sep 16 20:34:17 khem: did you have time to take a look at my patch updates ? I think http://patchwork.openembedded.org/patch/2957/ and 29{58,59,60} addressed all of your comments Sep 16 20:47:30 * kgilmer can't seem to clone the git repo Sep 16 20:47:42 'connection refused' it says Sep 16 20:47:44 o mean Sep 16 20:48:30 kgilmer works for somes and for others not Sep 16 20:48:34 kgilmer: I've had that same issue for the past 24 hours Sep 16 20:48:37 or its maintain now Sep 16 20:48:57 git fetch works here Sep 16 20:49:05 you have commit access? Sep 16 20:49:11 woglinde: it fails 100% of the time for me -- both pulls and clones Sep 16 20:49:14 maybe just ro is bad? Sep 16 20:49:26 Crofton: no commit access Sep 16 20:49:27 Crofton yeah Sep 16 20:49:41 hm pulling with rw works here Sep 16 20:49:55 ka6sox, did some work on it yesterday Sep 16 20:49:59 well the machine Sep 16 20:50:02 it fails with a machine I have in California, and another in B.C. Sep 16 20:50:35 in the past i've had intermittant failures but those were a different error, and it timed out for awhile. in this case it return immediately with 'connection refused'. Sep 16 20:51:10 resolving git.openembedded.org to 140.211.169.165 Sep 16 20:51:33 yeah Sep 16 20:51:42 ka6sox-work, ping Sep 16 20:51:55 woglinde, is it possible with existing openjdk recipes to just build zero...without building llvm? Sep 16 20:58:04 sakoman: strange, thought only the ro stuff would fail, pull worked for me both at home and at work Sep 16 20:58:37 ka6sox or whoever can do something about it: CIA could use a kick too Sep 16 20:59:19 git-daemon does not run on melo Sep 16 20:59:26 eFfeM: he knows about CIA Sep 16 21:02:03 ah ok, didn't know that Sep 16 21:02:15 ka6sox thanks again for your work! greatly appreciated Sep 16 21:04:01 kgilmer no Sep 16 21:04:41 kgilmer whats wrong with llvm? Sep 16 21:05:10 anyway, calling it a day, fresh build just started so i can get some sleep :-) have fun everyone! Sep 16 21:05:18 nite effem Sep 16 21:05:31 will go soon to Sep 16 21:05:39 waiting for kgilmer to answer Sep 16 21:09:42 hm Sep 16 21:09:43 okay Sep 16 21:09:48 good nite Sep 16 21:13:18 is there something wrong with the git repo? Sep 16 21:13:25 I'm getting connection refused trying to clone from it Sep 16 21:29:57 khem, do you know anything about the gcc atomic intrinsic code? Sep 16 22:10:10 rwxr-x---: hey, git daemon is not running. i think the admin is rsyncing the machine Sep 16 22:12:01 ok Sep 16 22:12:12 i managed to pull from the mirror Sep 16 22:17:49 i'm looking through the docs, and I'm trying to figure out the difference between your usage of OEDIR and OEBASE Sep 16 22:17:52 what's the difference? Sep 16 22:23:26 rwxr-x---: I am not sure which docs you follow. OEBASE sounds like BBCollections. Sep 16 22:24:42 Crofton|work: yes sure Sep 16 22:25:19 rwxr-x---: should we make you --------- Sep 16 22:27:30 khem, I am not impressed with the code the use :) Sep 16 22:27:36 at least for armv7 Sep 16 22:28:06 eglibc will use same code that is in glibc? Sep 16 22:28:20 urg Sep 16 22:28:26 the code is in gcc Sep 16 22:28:30 I am an idiot Sep 16 22:32:44 http://docs.openembedded.org/usermanual/usermanual.html Sep 16 22:32:53 trying to follow those docs for setting up a new architecture Sep 16 22:33:47 rwxr-x---: I am afraid I have some network trouble. sorry. Sep 16 22:33:47 starts getting fuzzy right from "Configuring OpenEmbedded" Sep 16 22:34:00 ok Sep 16 22:34:45 by the time I get to "Local ConfiguratioN" Sep 16 22:34:50 rwxr-x---: loaded, OEBASE is the directory that contains your openembedded/ and your bitbake/ directory Sep 16 22:34:56 the instructions seem to vacillate between editing multiple files Sep 16 22:34:58 ok Sep 16 22:35:24 zecke: is that the same as OEDIR? Sep 16 22:35:53 rwxr-x---: http://wiki.openembedded.net/index.php/Getting_started, might be better Sep 16 22:35:57 rwxr-x---: let me look Sep 16 22:35:59 let me rephrase the code in gcc for intrinsics on arm is dreadful Sep 16 22:36:22 Crofton: I think its good why do you think its bad Sep 16 22:36:27 well Sep 16 22:36:31 hang on Sep 16 22:36:57 actually it is not so much asm that might be a bit of overhead Sep 16 22:37:15 I need to perform some extra configuration for a set of pre-built cross compile tools Sep 16 22:37:38 just trying to figure out how to do it Sep 16 22:37:45 khem, this is how we fixed one case in boost Sep 16 22:37:46 http://pastebin.com/GvZDwWQT Sep 16 22:37:47 doesn't seem to be covered in the Getting Started guide Sep 16 22:38:03 rwxr-x---: ah okay, yeah OEDIR==OEBASE in that regard.. the documentation is not very consistent (patches welcome) Sep 16 22:38:23 rwxr-x---: ah, it is best to start with out it (let OE build the toolchain to at least have some success) Sep 16 22:38:25 this is disasm of a test program Sep 16 22:38:27 http://pastebin.com/3vi4kQPV Sep 16 22:38:37 the code seems generic Sep 16 22:38:58 I am not certain it guarantees atimic operations Sep 16 22:39:03 just that things compile Sep 16 22:39:04 my toolchain isn't upstream yet. Sep 16 22:39:08 this is a new architecture Sep 16 22:39:14 should work on uni processors Sep 16 22:39:35 my arm asm is weak though Sep 16 22:39:37 Crofton|work: no they may not be atomic Sep 16 22:39:46 so they will break on A9 Sep 16 22:39:48 ok so next doc question... what's LOCALDIR for? Sep 16 22:39:56 rwxr-x---: I wrote the 'external' toolchain docu, bit it is quite outdated, you want to use 'external-toolchain' which no one has documented so far. you can browse the mailinglist archive and search for that Sep 16 22:40:06 doh Sep 16 22:40:26 rwxr-x---: export LOCALDIR=$PWD/secret-isv Sep 16 22:40:35 what's secret-isv? Sep 16 22:40:39 my tendency is to repatch boost for the armv7 case Sep 16 22:40:57 hopefully eFfeM does not delete that patchfile ... Sep 16 22:41:16 rwxr-x---: ISV == Independent Software Vendor Sep 16 22:41:24 ok need to run Sep 16 22:41:35 ah Sep 16 22:42:16 Crofton|work: heh Sep 16 22:42:40 that makes a little more sense now Sep 16 22:42:45 rwxr-x---: I wrote this section of the document in 2006 when I used a motorola toolchain to build binaries for the EZX platform, quite some stuff has happened. Sep 16 22:43:03 sounds like a situation similar to what I'm in Sep 16 22:43:33 rwxr-x---: it is best to start with OEs toolchain Sep 16 22:43:35 I need to find someone who really understands arm and see what is up Sep 16 22:43:43 rwxr-x---: in most cases glibc/gcc are just compatible to older stuff Sep 16 22:43:52 at any rate, they can be greatly improved for armv6 and up Sep 16 22:43:58 l8r Sep 16 22:44:08 OE doesn't know how to build my toolchain Sep 16 22:44:37 rwxr-x---: what arch aree yoiu adding Sep 16 22:45:11 Crofton|work: I think gcc should use different builtin for armv7 already Sep 16 22:45:18 which gcc are you using Sep 16 22:45:21 it's a new one Sep 16 22:45:49 3.4.6 Sep 16 22:46:00 oh Sep 16 22:46:03 sorry heh Sep 16 22:46:08 rwxr-x---: Does support exist in toolchain for the "new one " Sep 16 22:46:25 not externally, yet Sep 16 22:46:39 btw gcc question was for Crofton|work Sep 16 22:47:12 rwxr-x---: the documentation will mostly work, you will need to add more ASSUME_PROVIDED for toolchain bits. you can use bitbake -g to generate a dot graph and then you can add binutils/gcc and such to ASSUMED_PROVIDED,then OE will not build these Sep 16 22:47:46 and also machine and tubne* files Sep 16 22:47:54 tune* Sep 16 22:48:18 I will be posting adding MIPS64 soon Sep 16 22:48:23 you can have a look at that Sep 16 22:48:32 ok Sep 16 22:48:33 and see what all needs to be touiched Sep 16 22:48:38 to get a new arch Sep 16 22:48:44 into OE Sep 16 22:48:46 cool Sep 16 22:48:51 khem: did you ever see *glibc* make install trying to install to /usr/local and ignoring the install_target? Sep 16 22:48:54 that will probably be helpful Sep 16 22:49:16 zecke: interesting is it newer glibc Sep 16 22:49:32 rwxr-x---: and rename your nick :) Sep 16 22:49:36 I've also already got uClibc that I'd want to assume provided Sep 16 22:49:46 all the good nicks were take :() Sep 16 22:50:03 so if I use: ASSUME_PROVIDED += " virtual/libc " Sep 16 22:50:11 and then set in the local.conf the type to linux-uclibc Sep 16 22:50:18 should that basically do it? Sep 16 22:50:37 TARGET_OS = "linux-uclibc" Sep 16 22:50:39 external-toolchain as zecke pointed out is the way for you Sep 16 22:50:44 look at that Sep 16 22:50:56 khem: we need docs for that. :) Sep 16 22:51:11 rwxr-x---: if you use distro's like minimal or angstrom TARGET_OS is already set internally Sep 16 22:51:22 is external-toolchain a doc somewhere? Sep 16 22:51:26 zecke: code is best document :) Sep 16 22:51:28 oh Sep 16 22:51:34 ok I didn't know that was i nhere Sep 16 22:51:45 what DISTRO do you use Sep 16 22:51:53 I would suggest to use minimal-uclibc Sep 16 22:52:00 if your system is uclibc based Sep 16 22:52:07 khem: (glibc) it is angstrom's default... it might be more a change in latest GNU make Sep 16 22:52:15 yes, right now my system is uclibc Sep 16 22:52:18 based Sep 16 22:52:27 khem: (glibc) it is GNU make 3.82 Sep 16 22:52:31 not sure what you mean by what distro; I don't have a distro for this architecture yet Sep 16 22:52:35 zecke: yes gmake 3.82 has issues Sep 16 22:52:52 rwxr-x---: DISTRO and MACHINE drives OE Sep 16 22:53:06 I see Sep 16 22:53:12 I think I just want a minimal right now Sep 16 22:53:17 just to get things working Sep 16 22:53:26 rwxr-x---: and you can choose something thats there or you can make your own Sep 16 22:54:02 rwxr-x---: minimal uses eglibc and minimal-uclibc uses uclibc as system C library Sep 16 22:54:11 that sounds like what I want then Sep 16 22:56:01 ok so this is starting to make some sense Sep 16 22:56:50 I should point BBFILES in local.conf to external-toolchain's BB files in recipes/meta? Sep 16 22:57:02 no... hrm Sep 16 22:57:27 not sure how external-toolchain files are selected/used Sep 16 22:57:40 rwxr-x---: external-toolchain is a bbfile (recipes/meta) Sep 16 22:58:09 rwxr-x---: you will need to set PREFERRED_PROVIDER for the toolchain bits to external-toolchain, and use SDKPATH environment variable to point to your toolchain Sep 16 22:59:23 kergoth: ping Sep 16 23:33:14 ka6sox: hmm bitbake git pull is not working for me Sep 16 23:36:21 ka6sox: doesn't work for me either (for more than a day now): http://pastebin.com/JTFgpU1f Sep 17 00:11:22 grg: ping Sep 17 00:11:29 Jin^eLD, pong Sep 17 00:11:42 got another patch for you, fix for an out of mem crash Sep 17 00:12:12 btw why is opkg_list_upgradable_cmd() duplicating opkg_list_upgradable_packages() ? :) Sep 17 00:12:34 err... dunno... Sep 17 00:12:37 :) Sep 17 00:12:45 lemme looksee Sep 17 00:13:20 well one calls the callback, the other one does a printf instead ;) would kind of make sense if the _cmd one just used the callback Sep 17 00:13:25 anyway, http://www.deadlock.dhs.org/jin/listupgr_oom.patch Sep 17 00:13:47 I made them look more or less the same, sorry had no time for refactoring Sep 17 00:13:52 the line that really matters is + node = active_list_next(head, node)) { Sep 17 00:14:23 the "node =" is missing in opkg_list_upgradable_packages() so it all went terribly wrong then Sep 17 00:15:25 there's a lot of duplication between opkg.c and opkg_cmd.c. The opkg_cmd stuff would ideally just use the library interface. But the lib interface is not really good enough to convert everything Sep 17 00:15:39 I see Sep 17 00:16:23 I have the feeling that I am the first one using the lib interface at all ;) Sep 17 00:16:38 first one in a long time Sep 17 00:16:58 i cleaned it up a tiny bit to use a subset of its features Sep 17 00:17:00 well, so far it did what I needed, aside problems like above :) Sep 17 01:17:21 hi , any player that is faster then omapfbplay and uses & utilizes dsp ? Sep 17 02:17:59 good evening Sep 17 02:18:24 u too Sep 17 02:18:34 has anyone seen the "garbage" that TI's X-loader emits causes mincom to dive into conniptions? cutecom seems to work fine, but minicom goes nuts Sep 17 02:19:08 tzanger , sry i donno Sep 17 02:19:12 btw Sep 17 02:19:26 when we use bitbake , where is compile logs ? Sep 17 02:36:55 tzanger: it is the OMAP boot rom that emits "garbage". see the TRM for details on the format of the data it sends **** ENDING LOGGING AT Fri Sep 17 02:59:57 2010