[05:08] Good morning! [05:26] Laney: hmm, I'm getting dead.letter files on the adt control host, the MAILTO= in the cronjob doesn't seem to work [05:26] oh, comma separated! === maclin1 is now known as maclin [06:13] Hi [06:20] Laney: FYI, lcy01 is dead again, disabled [06:54] good morning [06:54] bonjour didrocks, ça va / [06:54] ? [06:55] pitti: ça va bien, et toi ? [06:55] didrocks: je vais très bien, merci ! [06:59] good morning desktopers [07:00] hey didrocks pitti [07:00] morning seb128 [07:01] bonjour seb128 ! [07:03] happyaron, hey, did you see that https://launchpad.net/ubuntu/+source/fcitx-qt5/1.0.4-1 failed to build on some archs? [07:05] seb128: yep, but there's another report saying wait for Qt5.5 to land [07:05] k [07:05] Mirv is working on that it seems [07:05] yep, should be a Qt problem, I looked at the build but finds few can be done [07:06] k [07:06] thanks [07:06] bonjour! [07:06] ça va? [07:07] :) [07:07] hey larsu! ça va bien, et toi ? [07:07] seb128: sehr gut danke! [07:15] larsu, notify-osd tests are grumpy [07:16] bonjour larsu :) [07:16] larsu, there is one trying to load an icon from ../icons, or your fix doesn't allow for relative paths [07:16] larsu, unsure if we should just declare that one stupid and change it? [07:16] bonjour didrocks! ça va? [07:16] ça va, et toi ? [07:17] seb128: ya, this is clearly stupid. notify-osd doesn't know the cwd of the process that calls it and can thus not resolve a relative path [07:17] hey larsu, wie gehts? [07:17] * larsu is amazed that this was possible before [07:18] seb128: want me to remove that one? [07:18] larsu, can you update https://code.launchpad.net/~larsu/notify-osd/1520667/+merge/278861 to drop the buggy test? [07:18] larsu, yes please [07:18] moin moin pitti! Wie gehts? [07:18] seb128: sure [07:18] danke [07:18] or fix the test to make the path absolute, or just a name so that it gets looked up in the standard dirs? [07:19] transmitting relative paths over d-bus is bogus indeed [07:19] larsu: supi, danke! Zeit fuer Fruehstuck, also Stollen :) [07:19] mhhh lecker! [07:20] right, that test is a i18n [07:20] so better to rather update to use an non relative path [07:22] in fact just not setting an icon looks should be enough? [07:23] "wait_a_little()" sigh [07:23] lol [07:23] what exactly is this test testring? [07:23] that the API works? [07:24] no idea, notify-osd... [07:29] ya, I'm removing the icon [07:29] * larsu makes some breakfast [07:41] ugh, why do the tests show notifications on my system?! [07:41] what is this even testing? [07:43] likely a bug [07:43] and since they are queuing they spam your notify-osd for a while [07:43] I killed it to flush the queue [07:44] ya, that's what's happening right now [07:44] "you should see a bubble" [07:44] if I don't see it, how do I know the test failed? [07:45] quality software! [07:45] let's just drop the set_icon lines and move to something else [07:45] * larsu just pushed a better patch [07:45] before you itch to rewrite notify-osd again :p [07:45] becaue it uses that in multiple locations [07:46] and I don't know if I'd "break" the tests if I removed all of the icons [07:46] please test ;) [07:46] I drop the 3 set_icons calls in that file and make check was green [07:46] k, let me try your change [07:52] larsu, yep, that makes make check work, thanks [07:53] yw! [07:58] d-feet ships translations but displays untranslated but seems nobody ever noticed [08:07] * didrocks just sent to u-devel the great default language selection email! [08:07] let's see how it goes :) [08:13] morning [08:15] good morning Trevinho [08:15] hi didrocks [08:16] hey Trevinho! [08:16] hi seb128! [08:17] seb128: you know I already need your powers ? :) [08:17] https://bugs.launchpad.net/ubuntu/+source/xorg-gtest/+bug/1521366 [08:17] Ubuntu bug 1521366 in xorg-gtest (Ubuntu) "Compiz fails to build tests with recent g++ and xorg-gtest" [High,In progress] [08:17] oh [08:17] It's actually probably a pkgconfig thing, though.. [08:27] Trevinho, thanks for forwarding the gtest one to Debian btw! [08:32] seb128: thank you [08:51] $GREETINGS [08:53] hey willcooke [08:53] hi willcooke [08:53] morning willcooke [08:54] sorry [08:54] $HEY $BOSS [08:54] I meant [08:54] :D [08:55] $GREETINGTOBOSS [08:55] seb128: the fact that even $HEY is a var might be risky... [08:56] all those ubound variables creates headaches here :p [08:56] you know... if you define HEY=badword... :-D [08:56] unbound* [08:57] Trevinho, security issue, can be used to fire somebody! [08:57] yeah :D [09:02] morning! [09:02] HEY=sudo BOSS="pay laney all the money"? [09:02] lol [09:03] morning rich Laney ! [09:03] hey Laney [09:03] Trevinho, can you forward that xorg-gtest change upstream as well? [09:03] hey seb128 and Trevinho! [09:03] seb128: I was thinking about... I got just lazy :-D [09:03] well don't >:( [09:04] or your colleagues have to maintain it forever >:( [09:04] seb128: I'll do it, I just didn't do that last night as it was already late [09:04] Laney: I love my colleagues [09:04] hey pitti [09:04] good morning Laney [09:04] lcy01> oh noes [09:04] sup didrocks [09:05] Laney: good good, I guess I just avoided catching another cold (feeling good today, no more things in the throat) [09:05] guess it was a good bet to not run yesterday :p [09:05] it's that clean Lyon air [09:05] how are you? [09:05] breathe deep! [09:05] Laney: right! :) [09:05] feeeeeeeeeeeeling good [09:06] http://www.airqualitynow.eu/fr/comparing_city_details.php?lyon [09:06] fingers are reminding me that I climbed last night [09:06] morning Laney! [09:06] haha [09:06] Laney: ahah, nice :p [09:07] YEEEEEEEEEEEEAAAAAAAAAHHHHHHHHHHHHH [09:07] I'm going to buy a big bag of rice [09:07] http://www.dpmclimbing.com/articles/view/way-iron-fist [09:07] hey larsu [09:08] what's happening? [09:08] did you win the scottish quiz? [09:08] Trevinho, https://launchpadlibrarian.net/228039152/xorg-gtest_0.7.1-0ubuntu2.debdiff [09:08] seb128: if you want to wait and fix that patch for me, I can forward it and you fix the DEP-3 [09:08] didn't go last night [09:08] do we need to escape the " at all? [09:08] was sleepy :) [09:08] Laney: how about you? [09:08] seb128: the thing is that we need it to output -D\"STRING\" ... [09:09] seb128: old pkg-config was doing it with just using \"STRING\"... [09:09] seb128: with recent one, we need to do \\\"STRING\\\" to get \"STRING\" [09:09] no idea why [09:09] I'm not good with this things [09:10] maybe didrocks or Laney or pitti know what's right there? [09:10] seb128: compare the output of [09:10] pkg-config --variable=CPPflags xorg-gtest [09:10] in xenial and wily/trusty.... [09:10] what changed? pkgconfig? [09:11] I guess so, although I don't see much in the changelog [09:12] seb128: hum, let me look at context [09:12] didrocks: so... compiz fails to build because the CPPflags it gets from xorg-gtest doesn't include the properly escaped string [09:12] I've fixed the xorg-gtest.pc, as pkg-config seems to act differently to what it used to be [09:13] It might even be a pkg-config bug, btw... [09:13] let me look at the .pc and .pc.in from this package, one sec [09:13] larsu: great! [09:14] trusty $ pkg-config --variable=CPPflags xorg-gtest [09:14] -I/usr/include -I/usr/include/xorg -I/usr/src/xorg-gtest -DDUMMY_CONF_PATH=\"/usr/share/xorg/gtest/dummy.conf\" -DLOGFILE_DIR=\"/tmp\" [09:14] xenial $ pkg-config --variable=CPPflags xorg-gtest [09:14] -I/usr/include -I/usr/include/xorg -I/usr/src/xorg-gtest -DDUMMY_CONF_PATH=/usr/share/xorg/gtest/dummy.conf -DLOGFILE_DIR=/tmp [09:15] cd why would you need escape in -DDUMMY_CONF_PATH for instance, you don't have blank space or any IFS [09:15] so it seems xenial pkg-config behavior is right [09:16] might be compiz which tries to unescape things? [09:16] * didrocks looks at compiz and how it got those variables [09:16] definition is... [09:16] CPPflags=-I${includedir} -I${includedir}/xorg -I${sourcedir} -DDUMMY_CONF_PATH=\"@DUMMY_CONF_PATH@\" -DLOGFILE_DIR=\"@LOGFILE_DIR@\" [09:16] didrocks: no... compiz is not wrong [09:16] why do you need the "" in there? [09:16] pkg-config gets the thing with no quotes [09:16] which sounds ok, you don't have spaces or any IFS [09:16] as I say [09:16] so "" or no quotes should be the same [09:16] (for compiz cmake) [09:16] didrocks: it's a compiler-defined string, it has to have quotes [09:17] I'd expect to get: -DDUMMY_CONF_PATH=\"/usr/share/xorg/gtest/dummy.conf\" (as it used to be) [09:17] not -DDUMMY_CONF_PATH=/usr/share/xorg/gtest/dummy.conf [09:18] or... -DDUMMY_CONF_PATH="/usr/share/xorg/gtest/dummy.conf" would be also wrong [09:18] as the shell would unquote that [09:18] but again, why do you need the quotes? [09:18] yeah, I still disagree you need quotes [09:18] because you quote strings in c [09:18] yeah [09:19] Laney: well, this is not c, this is a compiler call where you pass args [09:19] without quotes, they're just considered as define FOO bar, where bar is not a string, but a symbol [09:19] it's the same as #define FOO BAR [09:19] exactly [09:19] Trevinho: ah, you mean those variables are seeded directly in config.h? [09:20] no, they're passed to g++ [09:20] http://cgit.freedesktop.org/pkg-config/commit/?id=50c2867f4a6981e085c721d936c96f174f11f415 [09:20] is the behaviour change [09:20] so when you do g++ -DFOO=bar file.c it's like you do #define FOO bar in foo.c [09:20] git bisect ftw ;-) [09:20] seb128: lovely [09:21] So, if that would have defined as cflags, there were no reason for this [09:21] I guess I can forward the patch now [09:21] # expect cflags from whitespace [09:21] RESULT='-Dlala=misc -I/usr/white\ space/include -I$(top_builddir) -Iinclude\ dir -Iother\ include\ dir' [09:22] -> they expect -Dlala=misc [09:22] Laney, does the \\\ look right to you? [09:22] not -Dlala="misc" though [09:22] didrocks, right, because "it's likely the value will be used verbatim in shell command substitution" [09:22] which I'm not sure it's true [09:22] well it's not in the current situation it seems [09:23] seb128: I wonder if the issue isn't that one but rather than cmake pkgconfig shouldn't handle the case for non shell command substitution [09:23] didrocks: I was wondering the same, but it's not [09:23] didrocks: since if you launch pkg-config in shell manually it's different anyway [09:24] didrocks: also, cmake would change the whole CPPflag string, not only a part of it [09:27] ouch, there's no xorg-gtest project on fdo bugzilla [09:27] yeah, I couldn't find one either, report on xorg others? [09:27] I don't see the same behaviour as you btw [09:28] Laney: no? [09:28] Laney, which part is different? [09:28] seb128: I guess it's https://bugs.freedesktop.org/describecomponents.cgi?product=X.org%20integration%20tests [09:28] http://pastebin.ubuntu.com/13595672/ [09:29] that should break right? [09:29] yep [09:29] Laney: no untill you use that variable in code [09:29] ah, sorryu [09:29] hum, he's using it, no? [09:29] I didn't read the code :-D [09:29] only gcc side :D :D [09:29] ahah [09:30] it has quotes for me [09:30] Laney: you didn't print your output of pkg-config --variable=CPPflags xorg-gtest [09:32] well it works for me as well with wrong output [09:32] mh [09:34] So, one possibility is to make cmake to requote it... But it doesn't look wrong when generating the make files [09:35] http://pastebin.ubuntu.com/13595755/ [09:35] and this is the failure http://pastebin.ubuntu.com/13595768/ [09:36] the whole thing gets 'single quoted' http://pastebin.ubuntu.com/13595769/ [09:36] Laney: ah [09:36] yeah, only the escape is stripped [09:38] it's weird that just running it in shell it doesn't get the quotes out [09:38] so....... mh [09:38] who is putting them in>? [09:38] pkg-config itself? [09:39] in what? [09:39] ah, you mean inside the single quotes? [09:39] I guess it's pkg-config at this point... But it does it in a subshell not when calling it directly [09:40] http://pastebin.ubuntu.com/13595832/ [09:49] Im probably very late to the party but yesterday I configured my first user service units and timers with systemd and its pretty awesome ... [09:49] seb128: in the mean time we've a different solution https://bugs.freedesktop.org/show_bug.cgi?id=93191 [09:49] Freedesktop bug 93191 in * Other "xorg-gtest CPPflags string aren't properly stripped with new pkg-config" [Normal,New] [09:49] * Trevinho wrote wrong title [09:51] Trevinho, that seems similar to the patch you suggested on launchpad earlier? [09:51] seb128: it's the same [09:51] :) [09:51] k, I was wondering [09:51] seb128: Maybe there's a different solution, but in the mean time... [09:51] you tried to trick me with your "we've a different solution" :p [09:52] oh [09:52] I mean I've no clue why this output is different... [09:53] I think you should talk to a pkg-config upstream person [09:53] Laney, didrocks, wdyt? is that change harmless? or any better suggestion? [09:54] seb128: I don't really like we don't get to the end of it [09:54] I wonder if the issue isn't in cmake pkgconfig module [09:55] IIRC, this one doesn't exist and it's a copy from project to project [09:55] * didrocks looks [09:55] compiz is executing pkg-config itself [09:55] yeah, me neither, I just don't know how things are supposed to be escaped, so unsure what is right/wrong there [09:55] Laney: compiz cmakerie you meant? [09:55] yes [09:55] it's there yes [09:56] but I didn't see anything wrong [09:56] didrocks: as I posted the generated makefile have the proper strings [09:59] Trevinho: what you pasted has single quote, no? [09:59] Trevinho: and are not escaped [10:00] http://pastebin.ubuntu.com/13595755/ and http://pastebin.ubuntu.com/13595768/ [10:00] hey bzoltan_ [10:00] so, they makefile generated are not right? [10:00] hello willcooke [10:00] yay new notify-osd [10:00] Sweet5hark, bzoltan_ has a couple of questions about the Doc Viewer click that you might be able to help with, or point him to the man who knows more? [10:00] I was just going to ask about that [10:01] didrocks: what I meant is that cmake doesn't delete or add anything... It just puts there what it gets from pkg-config [10:01] So, Sweet5hark :) you are the dude who is knowledgeble about libreoffice click package... is that correct? [10:01] Trevinho: brb [10:01] Laney, trying to join the HO but failing...... bear with me [10:02] willcooke: oh shit [10:02] give me 2 [10:02] :) [10:02] bzoltan_: somewhat, I didnt do the click package itself. I just created the *.deb (libreoffice-vanilla) that is the base for the click. [10:03] Sweet5hark: What I am most interested is the container and click packaging of an X11 app. Do you know who I should talk to? [10:03] willcooke: that sounds suspiciously like Laney forgot, or just didn't want management there to ruin it with sensibility ;) [10:04] :) [10:05] * Laney puts pants on [10:05] Laney: oh okay now it all makes sense [10:05] bzoltan_: likely I cant help you much there (yet). I just threw a *.deb with a little deps as possible over the fence and then popey and friends picked it up from there ... [10:06] * popey wakes up [10:06] Laney: classy :) [10:06] bzoltan_, what you wanna know? [10:08] popey: I managed to reach the same point with the Ubuntu SDK IDE as Sweet5hark with the libreOffice... I have a single .deb package what has almost no dependencies. Now I would like to create a click package what comes with a container where i can put build tools and stuff... I want to create phone apps on the phone when the phone is in windowed mode with keyboard and external display. [10:09] bzoltan_, are these X11 apps that you expect to run with XMir? [10:09] * Sweet5hark lurks on bzoltan_ questions, as they likely are relevant for him soon too. [10:10] popey: yes, very much X11 [10:10] Then I can't help, you need bregma / kgunn / ctownsend [10:10] The libreoffice package we have isn't X, it runs as a command line tool, no X output [10:12] popey: ohh, I see. Thank you :) I go after these guys [10:12] sorry :) [10:28] Trevinho: that's not really the case [10:28] Trevinho: it's not passing which pkg-config is giving directly [10:28] as on my xenial, it's giving: [10:28] pkg-config --variable=CPPflags xorg-gtest.pc [10:28] -I/usr/include -I/usr/include/xorg -I/usr/src/xorg-gtest -DDUMMY_CONF_PATH=\"/usr/share/xorg/gtest/dummy.conf\" -DLOGFILE_DIR=\"/tmp\" [10:29] (I even retried on a clean freshly deboostrapped chroot to ensure it's not a local config issue) [10:29] so properly escaped [10:29] and after cmake -> the makefile will issue the call, giving unescape sequence [10:33] didrocks, how do you run that command? [10:33] seb128: the one above ^ ("pkg-config --variable=CPPflags xorg-gtest.pc") [10:33] oh, fun [10:34] $ pkg-config --variable=CPPflags /usr/lib/pkgconfig/xorg-gtest.pc [10:34] -I/usr/include -I/usr/include/xorg -I/usr/src/xorg-gtest -DDUMMY_CONF_PATH="/usr/share/xorg/gtest/dummy.conf" -DLOGFILE_DIR="/tmp" [10:34] on my xenial [10:34] interesting… [10:34] so, the only different is that I have a locally extracted xorg-gtest.pc [10:34] pkg-config --version? [10:34] let me install the distro one, but it's the same version [10:34] I gave it the system location one [10:35] 0.28 [10:36] that's why [10:36] yeah, it's in 0.29 [10:36] I'm unsure why I have that 0.28 version [10:36] I'm using stgraber launchpad download tool for my schroot [10:36] but it seems even on xenial, it downloaded a wily one! [10:36] (even if I specify a xenial version) [10:36] * didrocks wants his money back [10:37] haha [10:37] morning [10:38] hey andyrock [10:38] * didrocks ditch this and do a good old debootstrap [10:42] ok, seeing the unescape sequence now [10:43] the.... Problem is that with the "fix", the laney's example doesn't work anymore... mhmh [10:43] go talk to upstream [10:43] yeah, let me try Laney's example first [10:43] srsly [10:43] if that work here as well, I this the issue should really be discussed with pkg-config's upstream [10:44] as we are missing something in the way it's executed [10:44] they should have added a pkg-config --variable-unscaped instead of breaking old stuff though [10:44] I doubt they intended to break things [10:45] sure, also because this is a corner case [10:45] but, you know... Beakage is always right behind the corner [10:45] so probably if you explain the case to them you will get a sensible answer [10:45] breakage [10:45] Laney: yeah, confirming Laney's example. So, I think the issue is something in the way cmake/the generated make is invoking gcc [10:45] I guess yeah, giving a simple reproduceable case with our build tool [10:45] and talk to them [10:45] I would prefer reverting the pkg-config commit meanwhile than some hack elsewhere [10:45] +1 [10:46] but even that after talking to upstream or at least foundations [10:46] yeah [10:46] who does that? :😃 [10:47] you? :p [10:47] I know that Mithrandir does pkg-config stuff [10:48] and maintains the package [10:48] not sure about the guy who committed this change [10:50] * Laney screams at g-i test failure [10:50] YOU WERE SUPPOSED TO BE FIXED! [10:54] Laney: so you know people... ;) [10:54] dude! [10:54] didrocks: mh, ok... Well I've to drive back to Florence now, I'll do this later [10:54] be brave [10:54] I believe in you [10:55] * didrocks sees Laney's flags "Marcoooooooooooooooooooo" [10:55] :) [10:56] Trevinho, have a safe drive! [10:56] thanks [10:58] Trevinho, let us know when you have opened an upstream pkg-config report [10:58] Trevinho, how much is that blocking compiz/unity work? [10:58] I wonder if we should revert the pkgconfig commit meanwhile [10:59] or wait a bit for upstream to respond [10:59] I would wait, but that's just me [10:59] well actually I would talk to them directly instead of opening a bug first :) [11:00] do they have a channel? [11:00] no #pkgconfig or #pkg-config [11:00] on this IRC at least [11:01] don't know, I gave a nickname though [11:01] right, it's a bit weird to ping people on don't know though [11:01] oh well, Trevinho can figure it out [11:01] there's a mailing list too [11:01] Laney, and it's easier for us to wait, we don't do one thing, if you were coding on one project and couldn't build on the infra it would be annoying [11:02] well you can do what you think is best [11:03] I only am one person giving an opinion at the end of the day [11:03] right [11:03] I'm going to wait for Trevinho to be back and say how much it's blocking them [11:04] but I'm leaning toward just doing the revert to unblock, I hate to see unity work staling because they are fighting with distro issues rather than coding [11:04] * Laney shrugs [11:04] if this is a revert that we carry forever then that's not a good way out imo [11:06] I think I could be replaced by a bot at this point [11:06] right, I'm not saying to not forward it upstream/discuss with them [11:06] but yeah, we are going round [11:06] :) [11:07] let's move on, plan of action is to talk to upstream but I'm going to unblock unity work as well if the upstream discussion doesn't get immediate traction [11:08] that's up to you === hikiko is now known as hikiko|ln [11:12] larsu, I saw you talking with Company on #gtk+ yesterday about screenshot, was there an issue with your gtk change? [11:12] larsu, also did we fix the segfault issue it created? the SRU is verification-failed, unsure what to do with it [11:12] either we fix that changeset or drop it, the SRU is meant to fix the tooltips round corners as well which was supposed to be a SRU landing around release time [11:13] seb128: hm, that's unrelated, no? [11:13] what and what? [11:13] ya, Company thinks the patch was wrong (and has a good point) [11:13] we already have a plan [11:13] seb128: tooltips and screenshots [11:13] right [11:13] we just batched them in the same SRU [11:14] which failed verification because of the screenshot segfault [11:14] so we either need to unbatch now [11:14] oh, just tested the compiz build in sbuild and it worked [11:14] or to resolve the screenshot one so they can be verified [11:14] * Laney checks it runs the tests [11:15] seb128: ah, let me fix that [11:15] NO! [11:15] THE HORROR! [11:15]