=== asac_ is now known as asac [14:57] fta2: libxul-embedding == 1.9 .... libxul-embedding-1.9.1 == 1.9.1 [14:57] asac, ? [14:57] thats at least my understanding for now [14:57] 00:15 < fta> how can i force an app using libxul-embedding to use xul 1.9 when I have both xul 1.9 and 1.9.1 installed ? [14:58] nope, it doesn't work [14:58] fta: what doesnt work? [14:59] i build with the 1.9 sdk, so with libxul-embedding but once installed, if i have 1.9.1, this is what is used [14:59] bug 295490 [14:59] Launchpad bug 295490 in liferea "Liferea doesn't start with "Aborted" error." [Medium,Confirmed] https://launchpad.net/bugs/295490 [14:59] oh [14:59] fta: yeah. fix maxVersion [14:59] maxVersion in liferea ? [14:59] 1.9.* was wrong ... should have been 1.9.0.* [15:00] yeah [15:00] in the glue code [15:00] oh [15:00] cool [15:00] let me get it ;) [15:00] sigh. my network is soooooooooooo slow [15:00] 00:15 < fta> how can i force an app using libxul-embedding to use xul 1.9 when I have both xul 1.9 and 1.9.1 installed ? [15:00] oops [15:00] 1293B/s 21min21 [15:01] i think i have to reset my router [15:01] bb [15:02] ok lets try ;) [15:03] that worked ;) [15:03] 2s [15:04] fta: ./src/mozilla/mozsupport.cpp [15:04] http://paste.ubuntu.com/72380/ [15:04] ;) [15:04] maybe we should detect that during built [15:06] damn, this could have saved me some hours :( [15:07] fta: the other way would have been to make ls /etc/gre.d/ first spit out the 1.9.0 conf [15:07] and not the 1.9.1 conf [15:07] (which i expect is the case for you?) [15:07] hmm [15:07] fta: what does ls /etc/gre.d give you? [15:08] 1.9.0.4.system.conf 1.9.1b2pre.system.conf 2.0a1pre.system.conf libxul0d.conf [15:08] 1.9.1b2pre-restored.system.conf 1.9a8pre.system.conf.dpkg-bak firefox.conf [15:08] a bit strange. xulrunner just tests file-by-file and first match is used [15:09] so 1.9.0.4.system.conf should have been used. not sure if ls always gives the same order as 'C' dir listings [15:09] though [15:09] i thought it preferred the newest version, which would make sense [15:10] fta: from what i know it doesnt prefer newest [15:10] e.g. in the past i had to make 1.9.0 configs illegal to test with 1.9.1 [15:11] personally, I agree that newest would make more sense. we should look into fixing this [15:12] http://mxr.mozilla.org/mozilla/source/xpcom/glue/nsGREGlue.cpp#156 === asac_ is now known as asac [15:13] fta: http://mxr.mozilla.org/mozilla/source/xpcom/glue/nsGREGlue.cpp#527 [15:13] so its really just readdir and then match first that suites [15:14] in my case, 1.9.1 always wins [15:14] fta: please do a strace -eopen -f liferea [15:15] and check whether 1.9.0.4.conf is actually looked at [15:15] otherwise readdir doesnt return in alphabetic order, but inode or something [15:21] fta: #include [15:21] #include [15:21] fta: http://paste.ubuntu.com/72386/ [15:21] ;) [15:21] fta: try that code [15:21] and see which order it gives [15:21] here it seems like its "random" ;) [15:21] most likely your 1.9.1 preceeds the 1.9.0 stuff [15:22] http://paste.ubuntu.com/72387/ [15:22] ^^ thats my garbage struck dir [16:09] i'm fighting against a scons [16:15] asac: any idea about my bug? :( [16:23] armin76: which? [16:23] armin76: i think you never came back when i looked at a bug ;) [16:23] but now i forgot [16:24] yeah, had to run fsck on a reiser [16:24] gentoo bug 234110 [16:24] Bug 234110 on http://launchpad.net/bugs/234110 is private [16:24] meh [16:24] bug gentoo 234110 [16:24] stupid bot [16:25] http://bugs.gentoo.org/show_bug.cgi?id=234110 [16:25] heh [16:25] bugs.gentoo.org bug 234110 in Applications "www-client/mozilla-firefox-3.0* USE=xulrunner - hangs on first start" [Normal,New] [16:25] armin76: i would guess that those folks have a dirty install [16:25] why? [16:25] its one of those things that regularly pop up [16:25] e.g. nothing works -> reinstall works [16:25] -> compreg.dat in pkglibdir [16:25] or some other cruft [16:26] saw the strace? [16:26] armin76: they should attach a strace -eopen -f [16:26] no [16:26] only see a huge list of config options ;) [16:26] comment 19 [16:27] yeah still [16:27] its meaningless [16:27] when people strip strace they always post unimportant parts ;) [16:27] hmm [16:28] armin76: /usr/lib/xulrunner-1.9/components/libpipboot.so [16:28] whats taht? [16:28] debug build? [16:29] armin76: they should attach a complete strace -eopen -f firefox to the bug [16:30] dunno, i have that file, want it? :D [16:30] k, i'll say so, thanks [16:30] armin76: you ahve that file? [16:30] armin76: is that a debug build?` [16:30] nop [16:30] e [16:31] armin76: how do you install stuff? [16:31] make install? :) [16:32] armin76: well. i wouldnt trust make install to do the right thing for mozilla ;) [16:32] armin76: that file should be wrapped into libxul.so [16:32] and not be there [16:32] dpkg -L xulrunner-1.9 | grep pip [16:32] /usr/lib/xulrunner-1.9.0.4/chrome/pippki.manifest [16:32] /usr/lib/xulrunner-1.9.0.4/chrome/pippki.jar [16:32] /usr/lib/xulrunner-1.9.0.4/components/pipnss.xpt [16:32] /usr/lib/xulrunner-1.9.0.4/components/pippki.xpt [16:32] /usr/lib/xulrunner-1.9.0.4/components/pipboot.xpt [16:32] dpkg -L xulrunner-1.9 | grep so$ [16:32] /usr/lib/xulrunner-1.9.0.4/libpyxpcom.so [16:32] /usr/lib/xulrunner-1.9.0.4/libjemalloc.so [16:32] /usr/lib/xulrunner-1.9.0.4/libxpcom.so [16:32] /usr/lib/xulrunner-1.9.0.4/libmozjs.so [16:32] /usr/lib/xulrunner-1.9.0.4/python/xpcom/_xpcom.so [16:32] /usr/lib/xulrunner-1.9.0.4/components/libimgicon.so [16:32] /usr/lib/xulrunner-1.9.0.4/components/libdbusservice.so [16:32] /usr/lib/xulrunner-1.9.0.4/components/libpyloader.so [16:33] /usr/lib/xulrunner-1.9.0.4/libxul.so [16:33] /usr/lib/xulrunner-addons/plugins/libunixprintplugin.so [16:33] /usr/lib/xulrunner-addons/plugins/libnullplugin.so [16:33] /usr/lib/xulrunner-1.9.0.4/libnssckbi.so [16:33] sry ;) [16:33] armin76: which configure flags do you pass? [16:33] --enable-optimize [16:33] --disable-debug [16:33] is important [16:33] sec [16:34] fta: do you have a built tree lying around for xul 1.9? [16:34] http://rafb.net/p/5ar7KI50.html [16:35] asac, hm, not sure, i trashed a lot of my stuff to make some room for chromium [16:35] asac, why ? [16:36] fta: just wondere where libpipboot.so would be lying around in it ;) [16:36] building my own 1.9.0 stuff now [16:36] fta: e.g. gentoo ships /usr/lib/xulrunner-1.9/components/libpipboot.so [16:36] we dont have that [16:36] ;) [16:36] we have pipboot.xpt [16:36] which makes sense and i assume that we have it in libxul.so [16:36] armin76: do you have libxul.so at all? [16:37] armin76: urgh. you asked for a paste of the output? ... thats really really long ;) ... attaching would have been more readable i guess [16:38] asac, i ship components/libpipboot.so in sm2 and tb3, not in xul1.9*/ff3* [16:39] i see i have it in my openkomodo 5 tree too [16:39] oh [16:39] --disable-libxul [16:39] armin76: ^^ why that? [16:39] fta: yeah. i think its a libxul component [16:39] do we need that? [16:39] so those that dont use libxul, still have it [16:40] fta: no :) [16:40] fta: gentoo is doing something a bit wrong imo [16:40] --with-default-mozilla-five-home=/usr/lib/xulrunner-1.9 [16:40] i dont think that that is still needed either [16:40] asac: not sure [16:41] armin76: anyway. lets wait for the strace [16:41] asac: yup, i have it [16:42] armin76: you have libxul.so ? [16:42] yup [16:42] /usr/lib/xulrunner-1.9/libxul.so [16:42] /usr/lib/xulrunner-1.9/sdk/lib/libxul.so [16:42] armin76: and also the .so files? [16:42] e.g. libpipboot.so [16:43] yup [16:43] booouh, that's not upstream scheme, it looks like ice* [16:43] armin76: what size does libxul.so have for you? [16:43] fta: huh? [16:44] /usr/lib/xulrunner-1.9/sdk/ [16:44] fta: i dont think that debian ships all components + libxul [16:44] fta: yeah ... a bunch of distros do it that way [16:44] asac: 132k [16:45] fta: its a bit of a shame and alrewady caused discussion when people kept using -rpath [16:45] i'm facing the same problem with chromium [16:45] armin76: yeah thats tiny. a [16:45] -rw-r--r-- 1 root root 18201928 2008-11-12 19:49 /usr/lib/xulrunner-1.9.0.4/libxul.so [16:45] armin76: but is in line with the --disable-libxul thing [16:46] not sure why there is libxul.so at all when you built that way [16:46] armin76: but its definitly not supported upstream [16:46] i'm not sure where does the --disable-libxul thing comes... [16:46] fta: what do chromium folks do? [16:46] oh, i see it [16:47] armin76: have you looked at what we pass as configure flags? [16:47] yup, in fact i think i copied it from there :P [16:47] asac, so far, nothing. binaries are just in the objdir and all libs are static, i'm trying to change that [16:47] armin76: oh ;) [16:48] fta: does upstream want to provide ABI tracked libs? [16:48] otherwise linking them into main binary makes a bit of sense [16:48] good question [16:48] the binary is huge [16:48] > 250MB [16:48] fta: is that a problem? [16:48] (on its own ;)) [16:49] fta: i think we would like to keep v8 in a system lib [16:49] well, no, in the sense that it runs [16:50] asac, i already landed some system libs: http://codereview.chromium.org/10626 [16:50] but most create regressions [16:52] fta: why are your comments empty? [16:53] asac, because i just updated the patch [16:53] ok [16:53] i like that site, it's nice to do patch reviewing [16:54] imho, far better than bugzilla [16:54] yean [16:54] google splits bugs and patches [16:56] bugzilla should learn from this ;) [16:56] [reed], ^^ [16:56] mconnor, ^^ [16:59] armin76: what does gentoo do on unemerge? does it track files installed by make install? [17:02] asac: it uses a sandbox system, so it keeps track of everything merged to the real system [17:03] armin76: do you ship .autoreg? [17:05] asac: http://rafb.net/p/RMgPlv30.html [17:06] damn, patching those scons files is too much python in unknown territory. i need to step back a little bit. [17:06] /etc/gre.d/1.8.1.18.conf [17:06] ? [17:07] hrm...sec [17:07] ok [17:07] 1.9 is further down [17:07] armin76: where is ffox 3? [17:07] (only see mozilla-firefox) [17:08] asac: http://rafb.net/p/s48RV614.html and mozilla-firefox is ff3 [17:10] /usr/lib/mozilla-firefox/.autoreg [17:10] so does that get installed by make instakk [17:10] ? [17:11] sec [17:13] asac: yes [17:14] was checking if it got installed by some patch, but nope [17:15] seems impossible to build shared and static at the same time with scons :( [17:16] * asac wonders why scons is considered better than automake ;) [17:16] all in python [17:16] which is good.... if you like python [17:17] i dont see how that can qualify as better [17:17] i think they think that its better because of windows [17:17] well ... at least those that think that python build system is better ;) [17:18] scons brings in even more magic than make, debhelper and cdbs combined [17:18] i hate magic [17:18] well ... if its pure declarative its good [17:18] but thats not magic [17:19] i meant it's even more obscure [17:19] like maven for java [17:19] compared to ant [17:19] lol [17:19] http://crupp.de/?p=15 [17:19] rewad the first paragraph [17:19] the guy said he used autotools by ... [17:20] "I used hand-written makefiles...." [17:20] So I looked for alternatives, and decided to settle on SCons instead of GNU Autotools [17:20] he [17:20] "And now I decided to revert my decision. Why? Because SCons is far more complicated than it looks, while autoconf/automake is far more simple than it seems." [17:20] good guy [17:20] patching v8 build system promises to be a challenge [17:27] @time [17:27] Current time in Etc/UTC: November 15 2008, 17:27:00 - Next meeting: Americas Regional Board in 2 days [17:27] @time berlin [17:27] Current time in Europe/Berlin: November 15 2008, 18:27:03 - Next meeting: Americas Regional Board in 2 days [17:39] asac: porting ff to arm? :D [18:04] armin76, i monitor the arm builds, so far, xul just failed: http://launchpadlibrarian.net/19673125/buildlog_ubuntu-jaunty-armel.xulrunner-1.9_1.9.0.3%2Bnobinonly-0ubuntu1_FAILEDTOBUILD.txt.gz [18:05] -so far [18:06] i was about to say, so far, xul is waiting for gnome, but not anymore [18:07] fta: same as hppa did.... [18:08] failing on cookie stuff wrt warning treated as err [18:09] like for hppa, i don't have access to arm, so i don't plan to blindly try to fix it [18:11] may want to file a bug, then :) [18:12] fta: or filter -Werror :) [18:13] no, not the good cure [18:14] in fact [18:14] mozilla bug 417345 [18:14] Mozilla bug 417345 in XPCOM "build broken on sparc" [Normal,Resolved: fixed] http://bugzilla.mozilla.org/show_bug.cgi?id=417345 [18:14] ah, now that i remember [18:15] i did a "patch" for that [18:15] mozilla bug 436133 [18:15] Mozilla bug 436133 in Networking: Cookies "Cookies build failure on hppa" [Normal,Resolved: fixed] http://bugzilla.mozilla.org/show_bug.cgi?id=436133 [18:16] guess arm needs to be there as well :) [18:21] asac: see the gentoo bug plz [18:22] probably but i have no idea how arm and armel differs in term of flags [18:23] * fta is going to lurk in #ubuntu-arm for a while [18:58] What's going on with Firefox 3.0.4 in any version of Ubuntu (specifically Hardy)? [19:15] omg [19:15] asac: OMG BUMB! [20:02] jdhore1: all ready. security team missed to push it on Thu and since we dont release on Fri we have to wait till Mon [20:02] ah [20:02] armin76: the strace -eopen -f would be helpful regardless of whether it works or not [20:03] armin76: you have fixes for the alignment issues? [20:03] we see them on arm now [20:03] (armel) [20:04] asac: see mozilla bug 436133 [20:04] Mozilla bug 436133 in Networking: Cookies "Cookies build failure on hppa" [Normal,Resolved: fixed] http://bugzilla.mozilla.org/show_bug.cgi?id=436133 [20:04] asac: i punted --disable-libxul and now it works according to the guy [20:05] armin76: hmm. [20:05] armin76: still . i think he has garbage in his pkglibdir [20:06] most likely he will run into issues sooner or later again [20:15] asac: http://rafb.net/p/PiQoTI26.html [20:15] -r0 is with --disable-libxul, -r1 is without [20:15] yeah [20:15] armin76: http://paste.ubuntu.com/72517/ [20:16] asac: i don't know about armel, you guys should ask if its safe to run -Wcast-align on arm or not [20:16] s/arm/armel [20:17] armin76: who would be able to tell that thats "safe" i ngeneral? [20:17] i mean if it was safe then its a gcc bug that it warns at all ;) [20:17] whoever maintains arm on ubuntu or debian [20:18] well. debian just did it i think [20:18] not sure if they seriously considered whether its safe or not ;) [20:18] look the patch i did [20:19] yeah [20:19] easy enough to do [20:19] INTEL_CC? [20:20] who supports that? ;) [20:20] hppa,ia64 and sparc give errors if stuff is unaligned [20:20] sure [20:20] i didn't know arm did as well [20:20] armin76: armel does apparently [20:20] armin76: http://launchpadlibrarian.net/19673125/buildlog_ubuntu-jaunty-armel.xulrunner-1.9_1.9.0.3%2Bnobinonly-0ubuntu1_FAILEDTOBUILD.txt.gz [20:20] yup [20:20] thats where i have the paste from [20:21] wonder if the code could be fixed to properly case [20:21] cast [20:22] you are better than me at that [20:23] the difference between debian and ubuntu is that debian isn't using -Werror [20:23] while ubuntu does [20:23] i remember that they had the same issues on armel when it popped up [20:23] i think they then just deliberately passed -Wno-error [20:23] not sure if its really a gcc default [20:25] http://mxr.mozilla.org/mozilla/source/xpcom/glue/nsVoidArray.h#193 [20:25] -CXXFLAGS += $(WARNINGS_AS_ERRORS) <- thats from debian [20:25] http://mxr.mozilla.org/mozilla/source/xpcom/glue/nsVoidArray.cpp#142 [20:25] armin76: thats in xulrunner right? [20:26] yes [20:27] they're just doing that on xulrunner-1.9.0.3/netwerk/cookie/src/Makefile.in [20:28] char mAutoBuf[sizeof(Impl) + (kAutoBufSize - 1) * sizeof(void*)]; [20:29] asac: and btw, sparc is still sigbusing [20:30] armin76: in cookie? [20:31] no... [20:31] * armin76 thinks asac has really bad memory [20:31] i can confirm that ;) [20:31] mozilla bug 448658 [20:31] Mozilla bug 448658 in Phishing Protection "nsUrlClassifierDBService has bad alignment, causes SIGBUS" [Critical,New] http://bugzilla.mozilla.org/show_bug.cgi?id=448658 [20:31] now i remember ;) [20:32] armin76: but dont remember everythign. didnt we fix that? [20:32] ;) [20:33] our gcc has -Wformat -Wformat-security by default now [20:33] https://wiki.ubuntu.com/CompilerFlags [20:34] asac: not too much, you made me try with some patch that just delayed the sigbus 45 mins [20:35] on sparc ? [20:35] yeah, see that bug [20:35] armin76: what was that patch? [20:36] did i just paste that or post it somewhere? [20:36] one like debian's [20:36] let me see the logs [20:42] asac: you pasted me a patch, and then i saw the debian bug, and you told me that was the patch you pasted [20:44] asac: then you got stuck in the canada thing, but you mainly said that with 45 mins you could live with it, i filed that bug at upstream's bugzilla and you didn't say anything after that [20:45] oh thats long time ;) [20:47] asac: then we talked about it at the end of september, but you didn't said anything new [20:49] i just asked you if you'd think 3.0.2 will sigbus, since it was released at that time [20:49] and it did, so... [20:49] asac: btw, bumb! [20:49] why you didn't bumb 3.0.4 yet!!!! [20:50] armin76: in jaunty -> my lazyness in holiday mode [20:50] everywhere else: [20:50] 1:02 < asac> jdhore1: all ready. security team missed to push it on Thu and since we dont release on Fri we have to wait till Mon [20:50] armin76: ^^ [20:50] armin76: the lines dont match anymore :( [20:51] heh [20:51] go fix the sigbus on sparc :P [20:52] looking ;) [20:52] armin76: is it still the same? [20:54] asac: considering that the file(nsurlclassifierdbservice.cpp) hasn't been changed since 25th august, i'd say yes [20:54] well, it got changed today [20:54] armin76: the lines dont match though [20:55] ah [20:55] the patch fails? [20:55] no ... the lines in the backtrace dont match anything sensible in the code ;) [20:56] at lesta the nsTArray line matches ;) [20:56] http://mxr.mozilla.org/mozilla/source/xpcom/glue/nsTArray.h#590 [20:57] okay, let me run it again [20:57] but i saw at the logs that you told me it didn't make any sense [20:57] yeah ;) [20:57] at least i found the line where the static cast happens [20:57] (i think) [20:58] http://mxr.mozilla.org/mozilla/source/xpcom/glue/nsTArray.h#182 [20:59] hmm. so what does new Type ... do? [20:59] (e.g. wihtout ()) [21:00] new (static_cast(e)) E; [21:00] what does that do :)? [21:00] you ask me? :P [21:00] without assigning it anywhere :/ [21:00] fta: ^^ ;) [21:01] http://mxr.mozilla.org/mozilla/source/xpcom/glue/nsTArray.h#178 [21:04] fta: any clue if the cast implicitly binds the constructed thing to "e" = [21:04] = [21:04] ? [21:05] nope, i fried my brain earlier today inside a SConstruct file [21:10] something more simple ;) .... cout ... iostream doesnt have that? [21:11] ah ... namespace ;) [21:16] * armin76 is upgrading to jaunty [21:32] armin76: dont know. if you have a up-to-date backtrace of sigbus please show [21:32] yeah [21:32] gdb doesn't seem to run fine on jaunty, trying intrepid now [21:50] armin76: heh. now our toolchain guy kicks me for fixing armel + xulrunner [21:51] funny thing is that our toolchain needs xulrunner to bootstrap ;) [21:51] you fixed something? :D [21:51]