**** BEGIN LOGGING AT Tue Sep 03 02:59:58 2013 Sep 03 07:26:40 good morning Sep 03 07:31:57 morning all Sep 03 07:38:03 hi RP, all Sep 03 08:01:09 morning all Sep 03 08:42:26 morning Sep 03 08:42:39 Anyone know about pkg_postinst_${PN} execution? Sep 03 08:43:58 pev: it's run post installation Sep 03 08:45:08 tf: My understanding is that it runs both on the dev machine and on the target - is there a simple way to provoke it to run on either as it looks like my rule in the recipe never gets called in either case and cant work out how to debug it... Sep 03 08:45:53 pev it runs on the target if it did not succeed on the host Sep 03 08:46:12 you test the ${D} var to tell if you are on target or host Sep 03 08:46:36 tf: Yep, I know about that, but it doesnt seem to get called at all Sep 03 08:47:30 I cloned the logic from : https://github.com/openembedded/meta-oe/blob/master/meta-oe/recipes-support/debianutils/debianutils_2.30.bb Sep 03 08:48:08 tf: However adding echos (and operations that change files) to both cases and I dont see either the echos or the results of the operations Sep 03 08:48:27 so wondering how to explicitly force the rule to execute to debug... Sep 03 08:51:26 pev: and where are you looking for the changes/output? Sep 03 08:52:38 tf: For echos, Initially in the output of bitbake -v when building the image, also under /tmp/log/* - for the changes I had the rule touch files in the build dir and in /etc on the target using unique names Sep 03 08:53:20 the output should be in the rootfs log for the image Sep 03 08:55:03 tf: Can I get bitbake to *just* build the rootfs alone to test this? Sep 03 08:55:20 build the image Sep 03 08:56:44 Dear all, does anyone know exactly which Linux version will be used in Yocto release 1.5? Sep 03 08:56:49 Yeah, I'm doing that but wondered if you could isolate the rootfs phase of that alone... Sep 03 08:57:56 uvan: 3.10 Sep 03 09:00:05 uvan: and 3.4 (LTS) Sep 03 09:00:23 ant_work: i want to know exactly is 3.10.9 or 3.10 Sep 03 09:00:54 normally th elast patchsets are applied Sep 03 09:01:16 so the 3.10.x of october probably Sep 03 09:01:40 uvan: most recent patches already merged say it is based on 3.10.9 Sep 03 09:01:49 tf: OK, figured that out and it seems I can provoke that ok now, so second question is that if I've built it in to an image that I'm putting on to the target, at what point should the postinst code execute on the device and where from? Sep 03 09:02:09 ant_work: thanks ant_work, bluelighting Sep 03 09:02:18 uvan: https://www.kernel.org/ Sep 03 09:03:03 yes ant_work Sep 03 09:03:23 I guess it could be 3.10.12 by the time Sep 03 09:03:59 because we are prepare to migrate Linux to 3.10 version to match with yocto release 1.5. so need the release info Sep 03 09:04:00 but really, this is uninfluent: the recipe is updated Sep 03 09:04:18 that's what matters ;) Sep 03 09:06:27 Ok, I'm confused now - so pkg_postinst runs on the host when building the rootfs. However it looks like the postinst scripts live on the target under /var/lib/opkg - but if the package is already in the image from image creation time, this doesnt look like it's ever run? Sep 03 09:06:53 is it only ever run on the target if you manually add via opkg? Sep 03 09:07:47 uvan: best person to talk to is Bruce Ashfield (zeddii) for definitive answers on the final version Sep 03 09:09:00 bluelightning: thanks, i will find him Sep 03 09:11:23 ...and it looks like postinsts should be run by S98run-postinsts ; except thats not there on my system for some reason, and it gets created by /var/lib/opkg/info/opkg.postinst which looks like a catch22 if it's not being run itself?? Sep 03 09:12:41 pev: you don't by chance have opkg installed without package-management in IMAGE_FEATURES? Sep 03 09:20:39 bluelightning: I don't *think* so, just trying to work out what my recipe picks in from other layers... Is there a way to easily show what it's set at for your environment? Sep 03 09:21:58 bluelightning, Hey, just FYI, I am gradually renaming all my recipes to the libX-perl convention, but! There definitively was a perl-module-X convention at _some_ time, may be long ago, because I see perl-module-X as dependencies for many of the libX-perl recipes on various git's. But they _are_ the old ones, to be fair. :) Sep 03 09:22:07 But like I said, it's _Just_ FYI. No complaints. Sep 03 09:23:18 pev: bitbake -e your-image-name is the best way to inspect variable values Sep 03 09:23:20 bluelightning, Heh, I finally got time to actually do it this week, less on my table and boss' is out (And pre-approved my doing this). Sep 03 09:23:41 Stygia: awesome, much appreciated :) Sep 03 09:24:18 bluelightning, How much does it matter to write .inc files that are generic and then bb files that require it? Is that strongly important, or is having just a recipe for the necessary version fine? Sep 03 09:24:54 bluelightning: Ah, that's useful, thanks! That gives me : IMAGE_FEATURES="debug-tweaks package-management" Sep 03 09:24:56 Stygia: for the perl distribution itself I think perl-module-xyz is used yes, but outside of there libxyz-perl is the convention Sep 03 09:25:11 pev: ok, so not the situation I was thinking of in that case Sep 03 09:25:21 Hm :-/ Sep 03 09:25:47 bluelightning, Well either way, I'll make everything be libX-perl and put it in meta-perl, yea? Sep 03 09:25:47 Stygia: just the recipe is fine; in the old days we had inc files shared by different versions but that's less important these days Sep 03 09:25:54 bluelightning, Alright, good. Sep 03 09:26:00 bluelightning, Because that'd be sort of a pain to do. :P Sep 03 09:26:40 bluelightning, Hmm. If something depends on libc6, is that worth mentioning as a DEPENDS, or is that pretty much implicit? Sep 03 09:26:53 bluelightning, Specifically a DEPEND and not RDEPEND Sep 03 09:26:55 Stygia: right, splitting out the inc after the fact would be painful :) Sep 03 09:27:25 Stygia: a dependency on libc is implicit unless INHIBIT_DEFAULT_DEPS has been set to "1" Sep 03 09:27:29 Stygia: depends, really Sep 03 09:27:48 Stygia: but if it only depends on libc, then that needs to be in DEPENDS, nothing is really implict Sep 03 09:28:03 Sooo. Sep 03 09:28:09 Can you guys fight it out and tell me the answer? :) Sep 03 09:28:28 Stygia: go with bluelightning Sep 03 09:28:50 tf, Alright, if you say so. Sep 03 09:28:56 tf, I have time to do it right now, though. Sep 03 09:29:11 And I'm not gonna rewrite this another time. So how it gets to be now is how I'll commit it. Sep 03 09:30:10 Stygia: if INHIBIT_DEFAULT_DEPS is not set and HOST_SYS != BUILD_SYS (i.e. cross-compiling) then "virtual/${TARGET_PREFIX}gcc virtual/${TARGET_PREFIX}compilerlibs virtual/libc" is added to DEPENDS (you can see this in base.bbclass) Sep 03 09:30:42 bluelightning, Well that does sound pretty definitive. Sep 03 09:30:54 And I'll say if someone inhibits the default deps it's his problem that stuff doesn't compile, yea? Sep 03 09:31:16 right, you would only set that if you know the compiler/libc aren't needed Sep 03 09:38:03 bluelightning: Ah - it appears the opkg S98run-postsinsts is a self-deleting script so it might be running and just failing to call my own packages postinst it seems! I've added debug to itself to see if that's what's happening... Sep 03 09:41:54 pev: postinst scripts that compete successfully at rootfs time don't re-run on the target Sep 03 09:57:38 rburton: I used this as a template : https://bitbucket.org/lokkju/openpli-oe/src/37f87a47a88d/packages/netkit-tftp/netkit-tftpd_0.17.bb Sep 03 09:58:19 rburton: it does an "exit 1" which I assumed would signal a non-completion on the host during rootfs so that wouldn't occur? Sep 03 09:59:01 indeed Sep 03 09:59:37 rburton: Although given that I can see it running on the host I'm almost tempted to say **** it and jsut hack it in as this is seeming to take a hell of a long time to just change a couple of lines in a daemons .conf file... :-D Sep 03 10:01:02 hi sameo Sep 03 10:01:23 rburton: Hey there Sep 03 10:05:31 OK, so the S98run-postinsts seems to run correctly (and then delete itself) so the next question is why doesn't that in turn call my packages postinst? All the script does in the end is invoke "opkg-cl configure" so why would this call / not call my packages postinst? Sep 03 10:06:33 it should... Sep 03 10:09:23 Hm....! Sep 03 10:10:05 rburton: I'm a bit out of my depth knowing nothing much about opkg - is there an easy way to test/debug this kind of thing? Sep 03 10:23:45 pev, Man opkg? Sep 03 10:24:52 opkg is really great Sep 03 10:25:02 it runs very fast actually Sep 03 10:25:09 * eren lately trying the new image with opkg feeds Sep 03 10:29:48 Isn't it a bit weird that the RPROVIDES in the perl recipe use the perl-module-X convention, whereas CPAN modules use libX-perl? Sep 03 10:30:28 It strikes me as confusing. Sep 03 10:32:00 well, it does at least let you know where the package comes from Sep 03 10:32:21 but I'm not sure if there is any other reason for the difference Sep 03 10:32:29 FWIW it's been that way since the OE-Classic days Sep 03 10:33:33 Hmm alright. Sep 03 10:33:46 Well it just looks weird to have perl-module-strict as a dependency along with libcarp-perl, for example. Sep 03 10:35:50 And hmm. Would using metacpan instead of cpan for "homepage" be frowned upon? Metacpan is where I get the metadata and it's essentially an improved cpan (website). Sep 03 10:37:52 bluelightning, Any opinion? Sep 03 10:38:52 Stygia: I don't see an issue with that; as long as the HOMEPAGE link gives an up-to-date location for info on the upstream project, that should be OK Sep 03 10:39:45 bluelightning, Fantastic then, and yup, that's what metacpan does. Sep 03 10:39:57 bluelightning, Because that's what I have everywhere... it would be painful to fix. :P Sep 03 10:39:59 Well... Sep 03 10:40:32 find . -type f -print0 | xargs -0 sed -i "s/:\/\/meta/:\/\//gi" Sep 03 10:40:35 Not painful. :P Sep 03 10:40:38 But anyway, fine. Sep 03 11:34:49 Question, is there any issue with UTF8 characters in recipes that aren't part of the normal English charset? Sep 03 11:35:15 The author of one of the CPAN thingimajigs I'm writing a recipe for is called Ævar Arnfjörð Bjarmason. Sep 03 11:35:37 Icelandic name I'm pretty sure, but is there any issue keeping it like that in the recipe? Sep 03 11:36:46 Could be Norwegian, but never mind that. Sep 03 11:47:35 I suppose it could introduce issues with maintaining it: Most people can't do æ ø å ä ö â and so on... Sep 03 11:48:24 Stygia: its only got to be written once Sep 03 11:49:23 rburton, Yup. But I take it, then, no issues? Sep 03 11:49:41 I'm sorta hoping you've got your stuff ported to python3, that has unicode magically working. Python2 is more difficult. Sep 03 11:51:24 Well I'm gonna push recipes with UTF8 encoding unless anyone says differently. Sep 03 11:53:22 Stygia: watch out on patchwork Sep 03 11:53:50 I remember issues with dev names Sep 03 11:54:05 (developers) Sep 03 11:54:22 ant_work, patchwork? Sep 03 11:54:39 (developers, developers, developers) Sep 03 11:55:51 ant_work, I'm not sure what you mean. Sep 03 11:56:32 http://patchwork.openembedded.org/ i guess Sep 03 11:57:15 zibri, Hmm right. Sep 03 11:57:24 ant_work, So.... what do you recommend? Sep 03 11:57:40 And does anyone here know if bb handles UTF8 properly, or if there are confirmed issues? Sep 03 11:57:44 try a test msg Sep 03 11:58:01 I'm pretty sure most keep finger off utf8 Sep 03 11:58:15 cowardly Sep 03 11:58:32 ant_work, Hmm what was it again... bb_something? For a print statement? Sep 03 11:59:14 bb.note ? Sep 03 11:59:26 ant_work, I don't think so, If I wanted to print some UTF8 chars in a do_install_prepend? Sep 03 11:59:38 ant_work, is bb.note a function? Doesn't look like it to me from naming. Sep 03 11:59:39 `grep --include '*.inc' --include '*.bb' -r '[ßåäöÅÄÖÆæØø]' meta` doesn't give a single hit :( Sep 03 11:59:53 even Angstrom gave up Sep 03 11:59:58 ;) Sep 03 12:00:21 Now I wonder if it's Angstrøm, Angström, or what. :P Sep 03 12:00:21 but we like pioneers :) Sep 03 12:00:22 stygia: it's a python function, but you can use bbnote Sep 03 12:00:33 bbnote from shell fragments, that is Sep 03 12:00:40 zibri, bbnote ah right, so bbnote "æøå" Sep 03 12:02:05 stygia: a combination of both btw, Ã…ngström :) Sep 03 12:02:27 zibri, Hah, fair enough. Sep 03 12:02:36 zibri, So that'd be... Norwegian or Icelandic, then. Sep 03 12:02:42 swedish. Sep 03 12:02:44 Once again I'm guessing icelandic, AFAIK the Norwegians. Sep 03 12:02:55 zibri, Huh. Swedes use Å? Weird. :P Sep 03 12:03:06 http://en.wikipedia.org/wiki/Anders_Jonas_Ã…ngström :) Sep 03 12:03:32 I should take more vacations, I came back to find that apparently someone finally figured out the mysterious and sporadic failures of PSEUDO_UNLOAD to work on some systems. Probably. Sep 03 12:03:42 zibri, Oh I believe you, I just thought Æ Ø Å was for us Danes and the Norwegians. Sep 03 12:03:44 ... Of course, now I don't know why it ever *did* work. Sep 03 12:03:56 And then I thought that the Swedish used Â Ô and à Sep 03 12:04:01 Or something like that. :P Sep 03 12:04:19 stygia: heh, the link was more "this guy is cool" Sep 03 12:04:44 stygia: heh, that looks more like charset issues ;) Sep 03 12:04:45 zibri, Ah cool, I thought he was a CPAN contributer. :P Sep 03 12:05:14 But then, AFAIK, I do remember correctly - Modern swedes probably don't use Å. Sep 03 12:05:35 uhm. yeah we do :) Sep 03 12:06:30 zibri, Huh alright. Sep 03 12:06:30 or maybe i'm just not modern enough :( Sep 03 12:06:50 zibri, Well fair enough! I just thought you used the variants with dots and lines and such, that's what I normally see in Swedish writings. :) Sep 03 12:07:26 zibri, But I'm thinking you as a Swede probably know better. :P Sep 03 12:08:33 I'd hope so anyway. Sep 03 13:00:54 How to add meta-oe to yocto? Sep 03 13:01:08 Anyone succedded before? Sep 03 13:02:04 vicky_, did you add it to your bblayers.conf? Sep 03 13:02:17 I have a question. Will meta-oeconflict the meta of yocto ? Sep 03 13:02:23 yes i added Sep 03 13:02:46 but i am facing a parse error of " No receipe for busybox" Sep 03 13:03:11 it shouldn't Sep 03 13:03:23 I afraid of my image will corrupt because of this? Sep 03 13:03:33 try changing the order in bblayers.conf Sep 03 13:03:44 vicky_: I had the same issue Sep 03 13:03:59 erbo, how did you fix it? Sep 03 13:04:03 I added meta-oe after poky layer definitions and got that error Sep 03 13:04:16 vicky_: you need to match the branch of meta-oe to the branch of poky/OE-Core you are using Sep 03 13:04:44 oh, yeah, forgot about that gotcha... it should be fresh in my memory. Sep 03 13:05:29 I was wondering, since I think we have fixed the bbappend in master Sep 03 13:05:51 k. Sep 03 13:06:07 what's the state of meta-raspberrypi btw? Has anyone tried that Sep 03 13:06:36 Will it affect of my compiled image? Sep 03 13:07:10 should i recompile again? Sep 03 13:10:04 hum... utils.bbclass' create_wrapper function creates an implicit bash dependency. i don't think that's intended. :-( Sep 03 13:22:24 zibri: isn't that only used on the build system binaries? Sep 03 13:23:32 I need a working toolchain that supports Qt 4.x on a beaglebone black, anyone have any recommendations? Sep 03 13:23:34 zibri, hmm. You said that you also have the issue with mails timing out due you our ISP's being idiotic. I'm using a VPN in the UK and I still have the same issue. How on earth do you actually submit patches? Sep 03 13:24:48 rp: we have an issue where depending on flex causes "Can't install flex-2.5.37-r0@mipsel_nf: no package provides /bin/bash" from smart Sep 03 13:25:13 flex is pulled in to target by depending on libpam it seems :/ Sep 03 13:26:13 stygia: either my mailing via a relay host (the isps usually provide these) or mailing from somewhere without this restriction. Sep 03 13:26:34 zibri: does the target system need the wrapper? Sep 03 13:26:43 usually its only for native Sep 03 13:27:21 not sure. it sets $M4 to ${bindir}/m4, i suspect not Sep 03 13:27:26 vicky_: shouldn't be needed Sep 03 13:29:34 also having problems with 'lld' command: http://pastebin.com/Lf2xhCPt when trying to examine executable. Sep 03 14:09:22 zibri: I'd suggest we make it native only Sep 03 14:09:33 native and nativesdk Sep 03 14:13:52 RP: any better way to install -klcc-cross besides -gcc-cross? http://tinyurl.com/o3wew4b Sep 03 14:20:09 zibri, Hey did you reply to me earlier? I fear my client timed out or something because there's a huge gap in talk on all channels. Sep 03 14:20:28 ant_work: what is wrong atm? Sep 03 14:22:02 iirc there is an hardcoded path to sysroot in the package Sep 03 14:22:59 well, sorry, there isn't any package Sep 03 14:23:20 it is just installed in sysroot Sep 03 14:23:33 as build dependency Sep 03 14:25:38 15:26:13 < zibri> stygia: either my mailing via a relay host (the isps usually provide these) or mailing from somewhere without this restriction. Sep 03 14:25:42 stygia: --^ Sep 03 14:26:05 zibri, Ah damn, nope I didn't get that. Suppose I could disconnect before I switch around between VPN's, heh. Sep 03 14:26:17 RP: basically any recipe inheriting klibc.bbclass -> DEPENDS =+ "klcc-cross" Sep 03 14:27:18 RP: feel free to abuse of the recipe to squash out unforgivable bugs like the sstate symlinks last time ;) Sep 03 14:35:30 zibri, And hmm right, I'm trying to set that up with my mail provider, but no luck... and seems like nobody replies to me at #postfix, #debian, #linux or really anywhere. I was afraid I'd been hellbanned or something. Sep 03 14:35:53 I don't think I'm nearly annoying enough to warrant that, though. I hope. Sep 03 14:37:43 Stygia: what's the problem with mails? Sep 03 14:38:17 I've set up a few generic mail servers, maybe I could help Sep 03 14:38:23 eren, Well. My ISP blocks port 25. I am trying to set up a relayhost with my mail provider, but it complains about TLS, although the main.cf file for postfix shows it as enabled and lists pem and cert files. Sep 03 14:38:34 eren, That'd be awesome. This is preventing me from posting patches. :/ Sep 03 14:38:57 Basically, I'm using postfix. git send-email ends up timing out, that's what I see in postqueue -p Sep 03 14:39:48 So I tried to set relayhost = [mail.provider.com]:587 and list the login info in /etc/ postfix/relay_passwd as per several tutorials online. Sep 03 14:40:21 But postqueue -p just shows "SASL authentication failed; server mail.PROVIDER.com[54.229.118.53] said: 538 auth not available without TLS " Sep 03 14:40:46 ldd on /usr/lib/postfix/smtpd shows it links to ssl and crypto, so it's compiled in. Sep 03 14:41:05 Config: http://pastebin.com/E13m9ayy, mail.log: http://pastebin.com/9fHasB2i, mail.info: http://pastebin.com/e4qU27RA, syslog: http://pastebin.com/PgLVbPg1(mail.err has nothing relevant - only my syntax errors) Sep 03 14:41:14 Stygia: oh, why do you need a mail server in the first place if you have a mail account from another provider? Sep 03 14:41:28 eren, How do you mean? I'm not trying to set up a server. Sep 03 14:41:49 eren, I just want to send emails with git send-email. Using postfix, I get errors. So as far as I could read I need to use a relay. Sep 03 14:42:25 Stygia: git send-email has option for setting up smtp account Sep 03 14:42:31 eren, If I'm not making sense I probably misunderstood something somewhere. Sep 03 14:43:05 Stygia: http://morefedora.blogspot.de/2009/02/configuring-git-send-email-to-use-gmail.html Sep 03 14:43:16 Stygia: change the smtp server information with your provider Sep 03 14:43:28 you probably know it Sep 03 14:43:45 when you set git configuration, git send-email will ask your password to log into your smtp server Sep 03 14:43:57 in this case, your mail account Sep 03 14:44:25 since you have "mail.provider.com", git send-mail will connect to your provider and send it :) Sep 03 14:44:38 you do not need a mail server to relay mails Sep 03 14:48:25 eren, Hah. Well... that works. :) Sep 03 14:48:34 eren, Awesome. I was doing something wrong all along, then. Sep 03 14:48:39 There's a patch for you all! Sep 03 14:49:38 And heh... Sep 03 14:49:42 Patch spam incoming. Sep 03 14:50:05 Stygia: no problem :) Sep 03 15:00:27 YPTM: Cristian joined Sep 03 15:01:41 YPTM: nitin is on the bridge Sep 03 15:01:55 YPTM: polk is here Sep 03 15:02:00 YPTM: We are about to start the Yocto Project Techincal Meeting (YPTM), this call is open to all Sep 03 15:02:01 Dial-in number: 1.972.995.7777 Sep 03 15:02:01 Participant passcode: 42001078 Sep 03 15:02:09 adamian is here Sep 03 15:02:13 If you have any issues feel free to join the call, this does not prevent other discussion here Sep 03 15:02:13 YPTM: belen joined Sep 03 15:02:15 cristiana is here Sep 03 15:02:24 YPTM: davest on the call Sep 03 15:02:26 YPTM: Corneliu joined Sep 03 15:02:39 YPTM: Jefro here Sep 03 15:03:01 YPTM: Saul is on Sep 03 15:03:05 YPTM: LaurentiuP joined Sep 03 15:03:06 YPTM: Paul Eggleton joined Sep 03 15:03:12 YPTM: Welcome to the technical meeting. Please let me know who's on the bridge Sep 03 15:03:13 YPTM: Tom Z here Sep 03 15:03:21 YPTM: Scott Rifenbark is on the call. Sep 03 15:03:32 YPTM: Richard is on the call Sep 03 15:03:33 YPTM: Michael here. Sep 03 15:03:46 YPTM: Björn is on the call Sep 03 15:03:56 YPTM: jzhang's on Sep 03 15:03:57 YPTM: pabigot listening in Sep 03 15:04:23 YPTM: Corneliu joined Sep 03 15:04:31 YPTM: pabigot is Peter Bigot ; t is silent (french) Sep 03 15:04:39 * simar YPTM: is on the call Sep 03 15:04:54 YPTM: Matthew joined Sep 03 15:04:57 Cristiana on the call Sep 03 15:05:01 YPTM: Bruce is on the call. Sep 03 15:05:26 YPTM: Any opens? Sep 03 15:06:07 Someone's small child has an open? Sep 03 15:06:24 sorry, thought I'd already muted Sep 03 15:06:37 we do welcome all contributions ;) Sep 03 15:06:42 BSDCat: (there was a ;-) implied there btw) Sep 03 15:06:43 :)) Sep 03 15:06:47 in case that wasn't clear Sep 03 15:07:07 :-) Sep 03 15:07:13 hm, YP on Lego Sep 03 15:07:25 mihai: :)) Sep 03 15:07:51 YPTM: Eren is here Sep 03 15:08:07 YPTM: 1.5 M4_rc2 qa report is not ready yet, but the current status can be found on https://wiki.yoctoproject.org/wiki/1.5_QA_Status Sep 03 15:09:30 ops? Sep 03 15:09:32 no Sep 03 15:09:45 interference? :) Sep 03 15:09:51 YPTM: Please mute your phones Sep 03 15:11:51 maybe the alien *is* talking! Sep 03 15:12:11 probably the guy at the NSA forgot to mute Sep 03 15:12:18 * zeddii nods Sep 03 15:12:19 using advanced audio compression Sep 03 15:15:04 any more information about security layer and related things? Sep 03 15:17:26 YPTM: I suspect someone *is* trying to talk but it's coming out garbled Sep 03 15:17:46 YPTM: probably a soft phone clipping packets like crazy Sep 03 15:36:52 seebs: you back? Sep 03 15:36:58 RP, An improvement of 750x. Nice! Sep 03 15:37:30 Somewhat back. Although the insomnia bug bit, so I'm pretty useless today. Sep 03 15:37:51 I did notice that someone finally found that !@#*!# pseudo bug. Sep 03 15:38:01 seebs: Ok, there was a patch to pseudo offered up just after you left. Sep 03 15:38:11 seebs: yes that's the one, it's pending your review. Sep 03 15:38:11 Oh, hey. I betcha that's pabigot, who's in-channel. Sep 03 15:38:24 seebs: yup Sep 03 15:38:25 yep. looking forward to the official patch ;-) Sep 03 15:38:54 I was just writing a response to it. Long-story short: I think we should adopt the given patch for now, and I am waffling as to what to do in the longer-term. Sep 03 15:39:00 hooray pabigot Sep 03 15:39:03 There is an interesting theoretical question. Sep 03 15:39:27 Oh, and pabigot, this makes you I think the second person ever to find a bug in pseudo before I found it. :) Sep 03 15:40:03 (Curious coincidence: You're "Peter A.", I'm "A. Peter". And computers and such being what they are, at least some computers record my name as "Peter A." Sep 03 15:40:04 ) Sep 03 15:43:27 ... aaand it appears that my work laptop has rejected the notion of network access. So I can't actually cause that message to be sent. Sep 03 15:45:39 gah, why are my -natives rebuilding, noooo Sep 03 15:45:54 Huh. I managed to update the bug, but the outgoing email got et. Sep 03 15:46:02 YPTM: thank you all for joining the call. Have a nice day/evening! Sep 03 15:46:16 kergoth: I have had great results from saying "stop it bitbake go home you are drunk". I mean, it doesn't really help, but it makes me feel better. Sep 03 15:46:25 hehe Sep 03 15:46:43 huh Sep 03 15:46:57 kergoth: bitbake-diffsigs -t ... Sep 03 15:47:13 fray, if you have access to the WR network, please pass on that I don't. I expect to get that fixed eventually... Sep 03 16:00:53 stygia: hum... a lot of the recipes you posted are core actually :/ Sep 03 16:01:43 (i'm assuming you're emil; right?) Sep 03 16:03:00 zibri, Correct. Sep 03 16:03:07 zibri, And hmm right. But hey, I have plenty that aren't. Sep 03 16:03:14 :) Sep 03 16:03:18 zibri, And that other guy said I can use corelist to check from now on... so I will. Sep 03 16:03:23 zibri, Heh I hope... I have 100+ recipes. :P Sep 03 16:03:27 They can't all be core. Or even mostly. Sep 03 16:03:36 that other guy is me :) Sep 03 16:03:40 Hah, alright. Sep 03 16:04:02 Well alright. No more patches without corelist! Don't worry. Sep 03 16:04:08 i would also recommend looking at sending your patches as a series of patches. Sep 03 16:04:10 Maybe I screw up but at least I try to learn when I do. Sep 03 16:04:15 zibri, How'd you mean? Sep 03 16:04:39 zibri, Maybe I should just go read some docs on git, heh.. :) Sep 03 16:04:49 stygia: heh... everybody makes mistakes :) Sep 03 16:05:27 stygia: you can see other patches being sent as e.g. "[PATCH 12/20] ..." Sep 03 16:05:59 where each patch in the series belongs to the same thread for mail clients supporting that Sep 03 16:05:59 zibri, Aah. So that's what I do to send a serious, just add X/total? Sep 03 16:06:08 *series Sep 03 16:06:20 zibri, AFK for a bit, sloppy burger. Sep 03 16:06:23 actually git format-patch supports that, and makes sure to set the correct mail headers as well Sep 03 16:06:42 (and by extension, git send-email as well) Sep 03 16:16:22 seebs: ah that good old "how can this fail?" -> "how can this work?!" transform Sep 03 16:17:19 Yeah. Sep 03 16:17:37 My favorite is when, the moment I *understand* why the code can't ever work, it stops working even in cases previously known to work. Sep 03 16:17:38 ... Sep 03 16:18:32 several options then: 1) drinking 2) tableflip 3) go home, try again tomorrow Sep 03 16:18:41 Apropos of nothing in particular, I note that it appears that the VPN on my work laptop has never been set up correctly, and that the reason I can't access anything is that it's never worked. I don't think I've touched that config in maybe six months except to update passwords when they expire. Sep 03 16:30:36 hi, is it possible with opkg to have signed packages? Sep 03 16:31:32 not as far as I know Sep 03 16:31:46 RPM is the only packaging format that YP supports that has signed package capabilities Sep 03 16:34:22 hmm, that is bad for secure embedded projects. Sep 03 16:34:52 oh, btw, debian does not support signing through _foo Sep 03 16:34:56 does support* Sep 03 16:34:59 erm Sep 03 16:35:03 but not more conveniently than that. Sep 03 16:35:09 like we did for Harmattan. Sep 03 16:35:10 I seem to recall that opkg does have some kind of GPG support Sep 03 16:35:17 I have never used it though Sep 03 16:35:29 I think security is inevitable nowadays. Sep 03 16:35:43 if opkg does not support it, it is out of the question with people secure embedded needs. Sep 03 16:35:55 having* Sep 03 16:36:01 correct Sep 03 16:36:10 correct refers to? Sep 03 16:36:14 package-level security is only relevant if you actually use packages in your target system Sep 03 16:36:22 sure, we do. Sep 03 16:36:23 which many don't Sep 03 16:36:31 erm, I would argue about that Sep 03 16:36:41 how else would you ship software update? Sep 03 16:36:47 reflashing everything? Sep 03 16:36:52 basically, yes Sep 03 16:37:05 now that is which "many don't" Sep 03 16:37:27 and many do. there are two approaches. neither is "right". Sep 03 16:37:51 I do not see the "many do" Sep 03 16:38:01 lpapp: something like http://lists.openmoko.org/pipermail/opkg-devel/2008-May/000063.html ? Sep 03 16:38:06 the customers I am familiar with primarily reflash images for upgrades.. some do packages, but it's still fairly rare. Those that do packages (and need them signed) almost always use RPM Sep 03 16:38:06 in fact all my app store stuff etc on the phones update the software at hand, not the whole device firmware. Sep 03 16:38:34 upgrading the firmware when you do not actually touch the firmware, just an end software... well, that is bizarr. Sep 03 16:38:47 lpapp: its just a different use case, nothing to argue over Sep 03 16:38:55 Number of devices - big, number of devices that do upgrades - small, number of devices that do 'package' or 'file' upgrades - miniscule Sep 03 16:39:07 JaMa: yeah, that, or X509 Sep 03 16:39:10 PKCS, etc Sep 03 16:39:19 local signature, and all those. Sep 03 16:39:32 rburton: different use case? Sep 03 16:39:41 we are discussing one use case. Sep 03 16:39:45 updating a software, not a firmware. Sep 03 16:41:07 JaMa: docs? Sep 03 16:42:06 JaMa: ha, you'd think i'd remember that work Sep 03 16:42:32 JaMa: http://lists.busybox.net/pipermail/buildroot/2013-April/071129.html Sep 03 16:42:40 no one seems to care about the changes. :( Sep 03 16:43:26 Paul Barker maintains opkg now Sep 03 16:43:30 try asking him Sep 03 16:43:42 does he use irc? Sep 03 16:44:56 you can use ML (opkg-devel@googlegroups.com) maybe someone else will reply or be also interested in it Sep 03 16:47:54 what would you say if someone asked why it is worth switching to dylan from denzil and whether it is worth it at all? Are there significant improvements in dylan other than few recipe updates? Sep 03 16:48:08 I heard that there minor buildsystem performance improvements, but nothing major. Sep 03 16:48:14 are* Sep 03 16:52:39 systemd integration and read-only-rootfs enhancements spring to mind Sep 03 16:53:24 what is read-only-rootfs enchancements? Sep 03 16:53:44 also, why is systemd a significant improvement? Sep 03 16:56:23 well, assuming you care about systemd that is Sep 03 16:56:29 I do not. Sep 03 16:56:46 what is good about systemd? Sep 03 16:56:50 so* Sep 03 16:56:58 it's now possible to boot with your rootfs mounted read-only, previously that wasn't supported out of the box Sep 03 16:57:11 again, something dependent on your use case Sep 03 16:57:32 what do you mean it was not supported out of the box... we have been using ro for ages with denzil Sep 03 16:58:44 we made quite a number of changes to allow postinstall scripts to run during do_rootfs where previously they would have to run on first boot Sep 03 16:59:58 in a minimal build it may have worked previously, not sato will work with a ro rootfs Sep 03 17:00:03 s/not/now/ Sep 03 17:00:13 lpapp: good for you, but for a lot of the cases we tested, things just broke because they assumed it was OK to write to places in the rootfs Sep 03 17:00:50 right, so nothing for us in there. Sep 03 17:00:58 then it is probably not worth switching to dylan altogether. Sep 03 17:01:04 I mean I do not see huge benefits for us. Sep 03 17:01:17 sure Sep 03 17:01:35 lpapp: being on something currently maintained isn't of value to you? Sep 03 17:01:48 bluelightning: is not. Sep 03 17:01:54 ok then Sep 03 17:02:00 in fact it is more dangereous as it is not stable. Sep 03 17:02:07 not as stable as an older version. Sep 03 17:02:09 that is simply untrue Sep 03 17:02:32 sure, it is true. Sep 03 17:03:39 I strongly disagree Sep 03 17:04:14 that does not make the fact untrue. Sep 03 17:04:33 that you assert it does not make it true either Sep 03 17:04:39 sure, it does. Sep 03 17:04:48 I believe we have tested our software more than you with this stack. Sep 03 17:07:00 the regular tests we perform coupled with the fact that many users are able to use dylan as a base for their current development is indicative enough to me that it is stable Sep 03 17:07:29 I fail to see how many unrelated users to us should be a factor for our testing and results. Sep 03 17:08:01 again, you are trying to comment out of your scope. Sep 03 17:08:08 if you want anecdotal evidence, it exists Sep 03 17:08:08 just like I would not comment for others. Sep 03 17:08:19 if you care not for it, by all means continue not to care Sep 03 17:08:21 lpapp: sure you do Sep 03 17:08:57 correct, I do not care about unrelated stuff. Sep 03 17:09:38 what's the typical use of ro rootfs? The first one seems to "just boot and never write" so that the flash/cf/card whatever medium lasts longer? Sep 03 17:10:37 demos, systems that you want to boot and "refresh from blank" on reboot.. Sep 03 17:10:59 stable 'upgrade' system images.. and and you said.. keep destructive media from dieing Sep 03 17:11:14 ro rootfs is common in embedded systems that need to be precisely the same at every new boot Sep 03 17:11:14 (RO is often faster to boot.. not always though) Sep 03 17:11:20 eren: we use that for safety. Sep 03 17:11:24 eren: end users cannot mess up Sep 03 17:11:26 and keep users from mucking things up... Sep 03 17:11:34 eren: or we cannot mess up either for demos. Sep 03 17:11:35 thanks Sep 03 17:11:54 eren: or for systems which are simply not meant to be writable. Sep 03 17:12:40 a lot of systems I've worked on have a large flash (set RO at boot).. with a specialized / serial / protected write mechanism for user data, settigns, etc.. but any temp files, etc should be wiped at power off Sep 03 17:12:40 I like the idea except when it comes to updating software on the fly. Sep 03 17:13:10 many of the systems I'm familiar with are dual partition.. bootloader remembers that last booted partition.. Sep 03 17:13:53 and you field upgrade the "other" partition, from the one you have booted.. then switch Sep 03 17:14:00 I think what would be interesting is semi-ro rootfs. Sep 03 17:14:08 i.e. the vendor can update it. Sep 03 17:14:36 fray: good way to update the system Sep 03 17:14:41 RO boot / RW app partition isn't that unusual in many places either Sep 03 17:14:52 that way if the app partition dies.. they can just clear it and the device keeps working Sep 03 17:15:02 it's 'inconvienent' but hopefully prevents a brick Sep 03 17:15:04 I am not sure that is a good way Sep 03 17:15:17 that still does not protect against vendor privileges and caps issues. Sep 03 17:15:26 I mean a user can just upgrade the software than on the writable. Sep 03 17:15:45 I would rather see one ro rootfs which is updatable by the vendor; hint, security mechanism, but not selinux. Sep 03 17:16:19 RW is from the system perspective.. the user can -never- log into the device w/o directly hacking it.. then all bets are off anyway Sep 03 17:16:50 now, that is called poor security Sep 03 17:17:06 and exactly the reason behind basics rooting, like how the kernel.org was compromised. Sep 03 17:17:06 consumer devices are more likely to have a RW.. since the life span of the device is much more limited (couple of years planned life) vs automotive which is almost always RO w/ monitoring settings, data store and apps.. (expected life is 17-20 years) Sep 03 17:17:45 if I hand someone with the ability and tools to hack a device, I don't care how security it is.. it'll fall.. Sep 03 17:18:03 user can "never" log into the device.... Sep 03 17:18:07 1) It is easy to hack Sep 03 17:18:19 the question is how do you prevent un-intended intrusions, how do you detect customer intrusions and do the 'right' thing.. (invalidate support, invalidate specific functionality, etc..) Sep 03 17:18:20 2) Even if it was not, it is easy to install apps from app store, website, etc. Sep 03 17:18:57 fray: by a security framework. Sep 03 17:19:02 what we did for Harmattan, for instance. Sep 03 17:20:01 swapping partitions will not increase the complexity for intrusion, unfortunately. Sep 03 17:20:11 it still remains the same complexity, how to defend that partition. Sep 03 17:20:13 swapping partitions has nothing to do w/ security.. Sep 03 17:20:18 it has to do with field upgrade reliability.. Sep 03 17:20:24 if one partition is mangled, you have another to fall back to Sep 03 17:20:27 well, that is the whole point of rootfs for many. Sep 03 17:20:31 ro* Sep 03 17:20:39 safety and security. Sep 03 17:21:00 RO is only a piece of the reliability, safety, security, etc.. you have to design the system as a whole to meet the objectives you have.. Sep 03 17:21:18 if the device is never internet connected, and isn't life critical.. then it can horribly insecure and it doesn't matter.. Sep 03 17:21:18 and I am saying, wapping will not help there much. Sep 03 17:21:22 s+ Sep 03 17:21:39 if the device has a 17-20 year projected life span, you need to make explicit hardware and software designs to meet that objective.. Sep 03 17:21:46 and there are a huge range of devices between that Sep 03 17:22:47 if it was 2010-2011, I would say libsmack could be integrated into some recipes-security or so Sep 03 17:23:55 anyone built successfully from a Fedora 19 host? I have someone who's using it as a host, and the rpmbuild in the sysroot is segfaulting building gcc-native looks like: http://paste.fedoraproject.org/36792/13782287/ Sep 03 17:24:10 walters: master? Sep 03 17:24:19 gcc 4.8? Sep 03 17:24:27 lpapp: dylan Sep 03 17:24:28 it was fixed for arch a while ago in master... unfortunately was not released. Sep 03 17:24:33 lpapp: yeah gcc 4.8 on the host Sep 03 17:24:37 there you go Sep 03 17:24:41 backport the patch or use master Sep 03 17:24:51 master of...oe-core? gcc? Sep 03 17:24:56 poky Sep 03 17:25:10 if I had to guess, something is wrong w/ the sse2 optimized strlen.. but I'd need to actually see the code that failed.. Sep 03 17:25:26 fray: already solved issue. :-) Sep 03 17:25:28 I'm not sure I have dylan easily avaialble Sep 03 17:26:27 the line is question: Sep 03 17:26:28 length += strlen(*av++) + 1; Sep 03 17:26:43 const char ** av = (*p).argv; Sep 03 17:26:43 while (count--) { Sep 03 17:26:43 /* add one for null termination */ Sep 03 17:26:43 length += strlen(*av++) + 1; Sep 03 17:26:47 thats the whole code block.. Sep 03 17:26:53 anyway.. it does look glibc related to me Sep 03 17:27:08 fray: if you wanna think about a resolved issue like that, please use pastebin, like paste.kde.org. Sep 03 17:27:09 walters, can you point me to an upstream bug so I can CC? Sep 03 17:29:38 lpapp: hmm...which recipe did the patch modify? Sep 03 17:30:01 walters: I would also google ... ;-) Sep 03 17:31:02 walters: http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?h=dylan&id=950f2e453a2bd31764e99eb09154768e0c5049a4 Sep 03 17:31:06 made it for you. Sep 03 17:31:22 i found that when searching the git log, but that seemed to be about gcc building gcc Sep 03 17:32:05 walters: yes, which is what you need. Sep 03 17:32:14 not sure how i would have known that -faggressive-loop-optimizations was also the source of the miscompiled rpmbuild code just from that Sep 03 17:32:39 I would suggest master for F19 Sep 03 17:32:45 it is unsupported Sep 03 17:32:57 and fixes might go in, but dylan was not a that great release for unsupported distributions. Sep 03 17:33:47 * lpapp wonders why this continuously recurring issues are not among the known issues for the given platform Sep 03 17:33:52 these* Sep 03 17:56:13 rp: btw, i've made a patch for the flex do_install_append, just verifying it... will send it tomorrow. hope i understood you correctly :) Sep 03 18:26:54 rburton: Re: x11vnc: Add PACKAGECONFIG for xfixes, xinerama, xrandr and xdamage, making them enabled by default is enough, right? Sep 03 18:49:52 JaMa: yeah, although a vnc server without fixes+damage is a suboptimal server Sep 03 18:50:14 rburton: ok Sep 03 18:51:47 fixed in branch Sep 03 18:55:51 rburton: script only compares RDEPENDS in package built after world and in clean tmpdir Sep 03 18:56:21 rburton: so if it was showing autoenabled connman dependency it had to be linking with something provided by connman package Sep 03 18:56:45 but the .pc doesn't link against anything... Sep 03 18:57:58 rburton: hmm looks like I've added connman just for completeness after reading configure.ac Sep 03 18:58:01 WARN: packages/armv5te-oe-linux-gnueabi/telepathy-mission-control/telepathy-mission-control/latest lost dependency on libnmglib upower Sep 03 18:58:35 makes sense Sep 03 18:59:03 oddly upstream fixed this today Sep 03 19:00:45 hehe Sep 03 19:01:00 so why do you guys use rpm5? Sep 03 19:01:14 * JaMa doesn't Sep 03 19:01:37 Jasper: last time i asked that the response was "i think it had to do with recommends, that rpm4 didn't/doesn't support" Sep 03 19:01:48 Jasper: then again that was an idle curiousity because i also don't use rpm Sep 03 19:02:02 nothing else in the world uses rpm5 though Sep 03 19:02:27 lpapp, btw, -O0 didn't fix the rpmbuild crash Sep 03 19:04:15 rpm5 is effectively unmaintained. Sep 03 19:04:36 Jasper: that is not true, jbj and others are maintaining it Sep 03 19:04:52 I haven't seen jbj do anything in years, though. Sep 03 19:07:29 Jasper: not sure what you mean. Sep 03 19:07:55 lpapp, after I recompiled rpmbuild with -O0, it still segfaulted when trying to run somewhere Sep 03 19:08:19 Jasper: sorry, but why are you telling this to me? Sep 03 19:08:38 Jasper: we could theoretically switch to dpkg; in the end it's just temporary data that gets put into an ostree repo, the packages are discarded Sep 03 19:08:40 lpapp, because I thought you were the one telling walters about the gcc bug above Sep 03 19:08:56 yeah, I suppose that works Sep 03 19:09:10 Jasper: yes, apply the patch or use master. Sep 03 19:09:29 lpapp, but I did -O0 -- shouldn't that also stop it from applying the broken optimization? Sep 03 19:10:13 I have no idea what -O0 means in this context. Sep 03 19:10:22 why don't you just follow the recommendation Sep 03 19:10:28 walters: use opkg, its the one true way ;) Sep 03 19:11:25 Compiler options won't retroactively change the C library; if you're linking against a host C library which has a bug you can trigger, changing compiler options will rarely affect that. Sep 03 19:11:27 walters: (dpkg is the least maintained backend, but considering you don't actually use packages on the target, why do you care?) Sep 03 19:11:57 We care because I can't manage to compile yocto -- rpmbuild keeps segfaulting. Sep 03 19:12:19 that's a good reason! Sep 03 19:12:22 Jasper: I think the first thing to do is follow suggestions. Sep 03 19:12:26 Apparently it was due to a bug in gcc 4.8, but even when I compile rpmbuild with -O0, it still happens. Sep 03 19:12:33 lpapp, walters suggested to me to use -O0 Sep 03 19:12:42 That's the suggestion I followed. What suggestion do you have? Sep 03 19:12:51 what I wrote to him to do. :) Sep 03 19:12:58 http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?h=dylan&id=950f2e453a2bd31764e99eb09154768e0c5049a4 Sep 03 19:13:19 if that does not solve either, you may need more changes from master. Sep 03 19:13:34 lpapp, but why should that matter if I'm compiling rpmbuild with -O0? Sep 03 19:13:44 Arch is pioneer in making other ancient distros' life easier at upgrade. Sep 03 19:13:58 Jasper: do not ask, just do. :) Sep 03 19:14:07 Jasper: smarter people than us already went through this... Sep 03 19:14:13 lpapp, OK, so how do I do? Sep 03 19:14:32 as I wrote several times, you can apply it or use master. Sep 03 19:14:56 master of what? poky? Sep 03 19:15:09 as I wrote that, too, yes. Sep 03 19:15:15 Keep in mind I'm not using poky from upstream, I'm using walters's fork of it. Sep 03 19:15:46 well good luck with that, then. Sep 03 19:15:58 Thanks! Sep 03 19:16:00 and ask walters about issues. Sep 03 19:16:09 I have been. Sep 03 19:16:19 no, you were telling me stuff Sep 03 19:16:24 do not tell me about a fork. :D Sep 03 19:16:33 as I have no idea what that contains, etc. Sep 03 19:17:27 lpapp, https://github.com/cgwalters/poky Sep 03 19:17:45 actually, should probably point you to the correct branch: https://github.com/cgwalters/poky/tree/gnomeostree-3.10-dylan Sep 03 19:17:48 Jasper: why are you sending this to me when I asked not to? Sep 03 19:18:12 walters: dylan? pah. it's all about dora now, keep up! Sep 03 19:18:24 Jasper: just for fun we could backport that patch...but i don't believe it will help Sep 03 19:19:32 walters: switch to opkg, you'll get faster builds Sep 03 19:19:42 (as well as it not segfaulting) Sep 03 19:20:00 rpm does not segfault for me. Sep 03 19:20:05 hm, actually, i think you still need to build rpm then, for the dep tools Sep 03 19:20:41 (and likely arch is even newer than F19) Sep 03 19:26:34 RP: yocot-1.4.2 <= typo in the tag Sep 03 19:26:46 bluelightning: ^ Sep 03 19:26:49 doh Sep 03 19:26:58 pidge: ^^ Sep 03 19:27:04 ;-) Sep 03 19:27:32 mario-goulart: thx by warn us :) sgw_ mario-goulart noticed and pinged me about it Sep 03 19:27:36 heh at least you can delete tags :) Sep 03 19:27:52 :-) Sep 03 19:27:58 thanks guys! Sep 03 19:29:13 JaMa: yeah, I saw the e-mail :) Sep 03 19:31:02 oh, gotta go Sep 03 19:31:07 see you tomorrow! Sep 03 19:47:54 fray, lpapp: I think we found it: http://pastebin.mozilla.org/2966396 Sep 03 19:48:13 rburton: unfortunately rebasing is quite painful for me because of the highly custom deployment backend Sep 03 19:48:44 on the positive side, i'm not shipping lots of little untested blobs with shell scripts run as root attached to client machines Sep 03 19:51:20 I'm really curious if that truely is undefined behavior Sep 03 19:54:05 walters: it'd be good if it could be a separated layer instead of a fork Sep 03 19:54:29 walters: so it'd allow for easy reuse of code and updating against new Yocto releases Sep 03 19:54:32 otavio: i know, i'd love to be there Sep 03 19:54:37 lol Sep 03 19:56:02 I'm still thinking this is a gcc bug Sep 03 19:56:11 should be an easy enough reproducer Sep 03 20:07:35 Since when is underflowing an unsigned integer undefined behavior? *signed* can blow up on overflow, but unsigned are pretty safe. Sep 03 20:10:41 If count has an integer type, and it's starting out non-negative, "while (count--) {...}" should execute {...} count times, and leave count set to -1 (if signed) or FOO_MAX (if unsigned). Pretty sure that's well-defined. Sep 03 20:17:13 it seems like we have an UTF-8 problem in mailman :( Sep 03 20:17:33 my fullname is not displayed correctly on the list archives Sep 03 20:21:10 walters: you are using a fork... Sep 03 20:21:23 walters: not sure anyone can comment much on it in here. Sep 03 20:21:59 eren: use ascii like me. Sep 03 20:22:13 the life has proven easier that way. Sep 03 20:22:24 for me, at least. Sep 03 20:23:46 lpapp: yocto archives are ok, there is a problem with oe lists Sep 03 20:25:47 eren: you will have more problems here and there in the internet world, I am sure. Sep 03 20:26:44 lpapp: I am not quite sure abotu that :) Sep 03 20:26:48 about* Sep 03 20:27:02 you are just having issues with a simple mailing list. Sep 03 20:27:16 there are a lot more complex scenarios out there. Sep 03 20:27:39 lpapp: for example? Sep 03 20:28:07 sgw_, pidge Going to start taking things offline now. Sep 03 20:28:37 walters, so, I'm wrong. Unsigned integer underflow is fully defined. Sep 03 20:28:50 eren: tons of sites. Sep 03 20:28:59 Jasper: yeah, i was reading about that too Sep 03 20:29:04 eren: anyway, why do you bother that much with it? It is not a biggie. Sep 03 20:29:51 halstead: perfect timing, I just finished a download! Sep 03 20:30:52 * sgw_ -> Everyone: Autobuilder, Mail, Downloads and Web services for the Yocto Project is going down for an hour or so, longer for the AutoBuilders. Sep 03 20:31:02 halstead: all yours! Sep 03 20:31:13 sgw_, Thanks. Sep 03 20:32:44 walters, I ran ostbuild again and it's not segfaulting anymore. Guess I got a lucky break/ Sep 03 20:32:45 ? Sep 03 20:32:58 sgw_: thanks. Sep 03 20:35:37 Jasper: hmm. Sep 03 20:35:55 WARNING: Failed to fetch URL http://downloads.sourceforge.net/bjam/boost_1_53_0.tar.bz2, attempting MIRRORS if available Sep 03 20:36:20 seems "bjam" disappeared off the internet Sep 03 20:36:51 Jasper: i thought nothing disappears from the nets Sep 03 20:37:09 i love how we have three build tools building each other now Sep 03 20:37:19 ostbuild runs bitbake runs bjam Sep 03 20:37:34 hehe Sep 03 20:37:55 somehow in a chain of elaborate wrappers for make, value is generated Sep 03 20:38:02 Jasper: the big bang of build tools Sep 03 20:40:23 heh value.. do you work for corporate marketing? :) Sep 03 20:41:12 Jasper: any chance you have it in a local downloads? You managed to hit just as we took our mirror offline for an hour or so. Sep 03 20:41:31 Jasper: anyways...dunno, we can just leave this one aside as a heisenbug for now, i'll probably upgrade my home workstation to f19 at some point and we'll see if it still exists Sep 03 20:42:50 OK. Sep 03 20:43:01 sgw_, seemed to continue Sep 03 20:44:01 Jasper: what version of oe-core/poky are you building Sep 03 20:46:27 sgw -- dylan based Sep 03 20:51:38 sgw_, cgwalters's fork of poky Sep 03 21:21:52 walters, Error: ldd not found in the base feeds (qemux86_64 x86_64 noarch any all). Sep 03 21:22:23 odd, let me see... Sep 03 21:40:10 hi, yoctoproject.org and git.yoctoproject.org seem to be down. Sep 03 21:40:43 it sure feels like bitbake's parsing times are slightly longer than they used to be, wonder if its a general thing, or if if something is broken with my VMs.. hmm Sep 03 21:42:50 well, i can't clone from git.yoctoproject.org , and i can't get there with browser too. Sep 03 21:43:22 it seems to be down from here also, yes Sep 03 21:45:30 planned outage was announced above by sgw_ / halstead Sep 03 21:46:28 ok Sep 03 21:48:11 ndec: kergoth: Right, we are just hitting the hour point, they are moving the servers in do different racks with more power, should be back by around 3:00-3:15 if all goes well. Sep 03 21:48:43 sgw_: no problem. i just didn't notice the announcement. Sep 03 21:48:57 take your time ;-) Sep 03 21:49:45 ah, cool Sep 03 21:57:49 Services are starting to come back up. mail, git, and wiki are back. Sep 03 22:30:25 yay, I had to use touch conf/sanity.conf ;) Sep 03 22:31:41 don't touch my sanity! Sep 03 22:32:08 :) Sep 03 22:32:13 sounds like someone hit a nerve... Sep 03 22:32:43 I didn't know the build couldn't start unless yoctoproject.org is up Sep 03 22:33:02 Is someone doing statistic with that data ? Sep 03 22:37:21 abelloni: being reported to the NSA Sep 03 22:38:33 oh right Sep 03 22:38:40 I forgot about that :) Sep 03 22:39:11 I called my machine config world_dominator.conf Sep 03 22:39:15 halstead: I know you are busy, just checking if bugzilla is supposted to be back up yet. Sep 03 22:43:42 sgw_, Not quite yet. It's up next. I'm having vlan issues. Sep 03 22:54:11 www.yoctoproject.org down? Sep 03 22:54:32 jkridner: yes, we are moving machine to a new rack at osl Sep 03 22:54:44 timeframe? Sep 03 22:55:13 any mirror for sources served there? Sep 03 22:55:52 should be back in the next 30 minutes or so. Sep 03 22:56:08 mirror would be kernel.org for released versions Sep 03 22:57:01 hum, boardzoo.com is also done Sep 03 22:57:12 down Sep 03 23:04:02 sgw_: I don't see pseudo source there Sep 03 23:04:53 jkridner: is the pseudo tarball there in downloads, this might be a pidge question Sep 03 23:05:35 downloads.yoctoproject.org is down Sep 03 23:05:42 jkridner: http://mirrors.kernel.org/yocto/pseudo/ Sep 03 23:05:58 jkridner: that should be what you're looking for? Sep 03 23:06:16 yup. just need to figure out how to setup mirrors now. Thanks! Sep 03 23:21:44 * mranostay spots a jkridner Sep 03 23:21:52 howdy mranostay Sep 03 23:22:43 jkridner: how is it going in kridner-world? Sep 03 23:23:24 playing with Yocto/OE, so that often will say how I'm doing. :) Sep 03 23:23:43 heading to NYC and Rome Maker Faires soon. Sep 03 23:24:04 mranostay: you headed to ELC-E? Sep 03 23:25:08 yeah Sep 03 23:25:45 presenting anything fun? Sep 03 23:26:30 http://embeddedlinuxconferenceeu2013.sched.org/event/735d49d3756da42953f76b3137697d77#.UiZwHT_s_cw Sep 03 23:29:54 mranostay: btw, I wanted to know what king of logic analyzer you use Sep 03 23:30:28 I'm using sigrok with a cheap saleae clone Sep 03 23:30:38 and I don't really get good results Sep 03 23:31:26 OLS Sep 03 23:31:41 just ordered a Saleae Sep 03 23:32:38 you plan to use it with fx2lafw ? Sep 03 23:32:45 probably Sep 03 23:33:02 ok, I'm interested in your results Sep 03 23:33:08 so am i :) Sep 03 23:33:26 going to be a mad dash to get this all working for ELC :) Sep 03 23:33:46 I also probably should try to put all the inputs I'm not using to the ground Sep 03 23:34:17 abelloni: for sure Sep 03 23:34:26 big problem on the OLS Sep 03 23:34:39 they will screw up the channels you are tracing Sep 03 23:34:44 yeahn I saw your beautiful device on g+ ;) Sep 03 23:36:12 * mranostay wonders why they didn't put weak pulldowns on the inputs Sep 03 23:36:43 using one buffer chip so can't Hi-Z the unused channels Sep 03 23:37:55 https://lh6.googleusercontent.com/-hLDEHaMiOHA/UX53M49Xb9I/AAAAAAAAETU/PvsOHly8d20/w671-h503-no/IMG_20130429_152811.jpg Sep 03 23:38:03 this is my setup ;) Sep 03 23:38:29 sgw_, Bugzilla and other services are back up. Sep 03 23:38:44 halstead: thanks! Sep 03 23:39:04 sgw_, Builders should be coming back soon. Sep 03 23:39:16 halstead: what happened? Sep 03 23:44:40 mranostay, The move took longer than expected because the vlan setup changed between the racks and I needed to reconfigure the switches. Sep 03 23:52:03 abelloni: going to ELC? Sep 03 23:55:51 yeah Sep 03 23:56:11 I'm not presenting anything though Sep 03 23:56:19 ndec: why are there two of you on G+? :) Sep 03 23:56:41 there were already enough people from Free Electrons talking ;) Sep 03 23:56:54 heh Sep 03 23:59:44 I was quite disappointed to see that it is only two days this year Sep 04 00:00:28 the end result is that you have more chances of overlapping interesting talks Sep 04 00:03:44 abelloni: yeah Sep 04 00:03:51 at least my talk is the first day Sep 04 00:04:22 and not the first of the day Sep 04 00:06:35 abelloni: yeah i see a few overlaps Sep 04 00:06:55 panto or Tartarus... hard choice Sep 04 01:23:15 sgw_, All the builders except 02 are ready. You could start a MUT now. Sep 04 02:32:23 hi all, i try to create a new images and got error: Error: locale-base-en-us not found in the base feeds (genericarmv8 aarch64 noarch any all) Sep 04 02:33:48 could anyone tell me how to include locale in this case? **** ENDING LOGGING AT Wed Sep 04 02:59:58 2013