[00:05] <bobby__> Updated FF4 browser, XULRunner 2.0 still won't upgrade
[00:10] <chrisccoulson> bobby__, you need to wait for the builds to next run
[00:10] <bobby__> :(, when will that be?
[00:11] <chrisccoulson> in the next few hours
[00:11] <bobby__> :(
[00:11] <bobby__> No, I'll be fine, really
[00:12] <bobby__> Anyway, thats fine, it still works :D
[00:12] <chrisccoulson> cool
[00:12] <bobby__> ah dang it, Flash is crashing now
[00:12] <chrisccoulson> these are daily builds, so things will break from time to time ;)
[00:13] <chrisccoulson> (and at this stage, whilst we're getting things ready for natty, expect things to breka frequently)
[00:13] <bobby__> Well, I'm using 10.2 Square preview... But its never crashed before this actually...
[00:15] <bobby__> But if it IS your fault, then... Well... Yeah...
[11:51] <chrisccoulson> hi asac, did you see https://code.launchpad.net/~chrisccoulson/ubufox/ff-4.0/+merge/40373 :)
[15:42] <BUGabundo> Unable to open URL
[15:42] <BUGabundo> Error launching "xdg-open 'http://forum.xda-developers.com/showthread.php?t=805327'": Failed to fork (Cannot allocate memory)
[15:51] <BUGabundo> fta: need a way to limit chromium ram... it mem leaks and kills my system :(
[15:52] <fta> BUGabundo, well, limit/ulimit but not sure it will work
[15:53] <fta> haven't tried that in ~15y
[15:53] <BUGabundo> ehe
[15:53] <BUGabundo> now running in GDB
[15:53] <BUGabundo> but that uses even more RAM :(
[15:54] <Dimmuxx> buy more ram ;)
[15:55] <BUGabundo> 4GB not enough to run ONE browser and 4 tabs?
[15:55] <BUGabundo> damn it
[15:55] <Dimmuxx> depends on the browser
[15:58] <micahg> chrisccoulson: are we planning on installing mozjs in /usr/lib or will it just be a separate package?
[15:59] <chrisccoulson> micahg - not sure. one idea from UDS was to install a versioned libmozjs in /usr/lib (separate package) and have a non-versioned libmozjs symlink in the usual place (so that our toolkit remains ABI compatible with the official builds)
[15:59] <chrisccoulson> we're really going to be under pressure to drop xulrunner off the CD this cycle
[16:05] <micahg> chrisccoulson: hmmm, I was hoping we could get an accessible webkit :-/
[16:17] <fta> http://people.ubuntu.com/~fta/popcon-20101109-pct.png
[16:18] <micahg> fta: congrats
[16:18] <fta> :)
[16:21] <fta> that's almost a million users for chromium
[17:01] <fta> with the rest of the pack: http://people.ubuntu.com/~fta/popcon-20101109-pct2.png
[17:50] <jcastro> fta: surely the numbers for epiphany are wrong?
[17:50] <jcastro> I mean, don't get me wrong, I like it, but I can't believe there's that many users?
[17:51] <fta> jcastro, i assume it was installed by default a long time ago or something
[17:51] <micahg> jcastro: depends which column in popcon he's using, popcon is historical
[17:51] <fta> installed
[17:51] <fta> http://paste.ubuntu.com/528798/
[17:52] <jcastro> fta: maybe gnome dev packages bring it in?
[17:52] <jcastro> either that or it's everyone's naughty browser. :p
[17:53] <BUGabundo> :o
[17:53] <BUGabundo> private browsing would be easier
[17:54] <fta> jcastro, it's falling fast. at this trend, in 4~6 months, chromium should eat it
[17:54] <jcastro> BUGabundo: not if you want to keep bookmarks/history/etc.
[17:55] <fta> and 12 months for konq
[17:55] <jcastro> not like I would know *whistles*
[17:55] <jcastro> fta: yeah it just seems like it would have long surpassed it by now
[18:00] <fta> jcastro, jdstrand: so much for debian/fedora and their all-system-libs browsers: http://spectralhole.blogspot.com/2010/08/why-you-dont-want-to-build-you-chromium.html
[18:01] <jdstrand> heh
[18:38] <fta> oh, nice (if you have a 7" ereader and want to read magazines): http://www.pdfscissors.com/
[21:22] <chrisccoulson_> micahg - was it you who mentioned about dropping the update-notifier hook?
[21:23] <chrisccoulson_> i'm sure somebody talked to me at UDS about it, but i can't remember ;)
[21:23] <micahg> chrisccoulson_: I think I might have mentioned it
[21:23] <micahg> chrisccoulson_: I saw you did it, but if ubufox goes away, we'll have issues
[21:23] <chrisccoulson_> i don't think it will go away just yet ;)
[21:24] <chrisccoulson_> the notification area will go away before ubufox ;)
[21:24] <micahg> chrisccoulson_: ok
[21:24] <chrisccoulson_> and i think we were pretty much the only people using that mechanism in update-notifier anyway
[21:24] <chrisccoulson_> and that dialog was hideous ;)
[21:24] <micahg> chrisccoulson_: are we looking at a consistent path as well this cycle?
[21:25] <micahg> then we can get rid of the restart notification completelu
[21:25] <micahg> *completely
[21:25] <chrisccoulson_> not sure yet, but we'd still need to keep it around anyway, as FF would still break if files are renamed etc
[21:26] <chrisccoulson_> so we'd at least need it for the major version upgrade case
[21:26] <micahg> chrisccoulson_: that shouldn't be happening during point releases
[21:26] <micahg> right
[21:27] <micahg> I would almost prefer having people close Firefox or deferring the update in those cases rather than having the restart notification
[21:27] <chrisccoulson_> well, the second case might happen anyway. that's something i talked about with mvo at UDS
[21:28] <chrisccoulson_> but we'd still want a way to tell users that they need to close firefox for the upgrade to complete
[21:28] <micahg> chrisccoulson_: the same way any other package needs input on updates? (debconf)
[21:28] <chrisccoulson_> no debconf, that's horrible ;)
[21:29] <chrisccoulson_> mvo suggested adding a new field to debian/control that would tell u-m to defer unpacking the package
[21:29] <chrisccoulson_> we'd then have some magic in our launcher script to trigger the upgrade, and tell the user to restart the browser when an update is ready
[21:29] <chrisccoulson_> then the update would happen as the browser restarts
[21:30] <chrisccoulson_> that's only one idea though
[21:30] <micahg> chrisccoulson_: that would be cool and mirror upstream behaviour
[21:30] <chrisccoulson_> i need to talk to mvo when i get the chance though to flesh out the exact details
[21:30] <chrisccoulson_> but i think something like that could work
[21:30] <micahg> chrisccoulson_: the only issue I can think of is policykit access to the package installation
[21:31] <chrisccoulson_> that's what aptdaemon is for :)
[21:31] <chrisccoulson_> oh, yeah, actually
[21:31] <chrisccoulson_> i see what you mean
[21:32] <chrisccoulson_> we wouldn't want a second auth dialog to appear
[21:32] <micahg> chrisccoulson_: right
[21:33] <BUGabundo> yet another crash
[21:33] <BUGabundo> stupid chromium mem leaks :(
[21:34] <fta> BUGabundo, you might want to have a look at your mapping file to see what's leaking
[21:34] <BUGabundo> how?
[21:34] <fta> in the apport report
[21:35] <BUGabundo> its a system freeze
[21:35] <BUGabundo> nothing logged
[21:35] <BUGabundo> most of the times not even chromium restore
[21:35] <fta> hm
[21:35] <fta> anything relevant in dmesg?
[21:36] <BUGabundo> $ pastebinit /var/log/dmesg.0
[21:36] <BUGabundo> http://paste.ubuntu.com/528925/
[21:36] <BUGabundo> nothing I see
[21:37] <BUGabundo> nothing in syslog.1 either
[21:39] <BUGabundo> ill tee gdb to file or something
[21:39] <BUGabundo> how do I make chromium more verbose again?
[21:42] <fta> do you have a bug for that in lp?
[21:42] <fta> or upstream?
[21:45] <BUGabundo> not yet
[21:45] <BUGabundo> nothing to report
[21:46] <BUGabundo> not even reproducivle
[21:46] <BUGabundo> sometimes with no change in pages it just goes crazy
[21:46] <BUGabundo> but most of the times its when openning a new one
[21:46] <BUGabundo> tried to disable a few addons, but nothing changed
[21:47] <fta> do you even know which process goes crazy?
[21:47] <fta> about:memory
[21:47] <BUGabundo> I can't do anyting on my laptop
[21:47] <BUGabundo> I'm forced to power cycle
[21:47] <BUGabundo> it uses all my ram
[21:47] <BUGabundo> no swap (ssd)
[21:48] <BUGabundo> REISUB and reboot
[21:48] <BUGabundo> 1 window, 5 tabs
[21:48] <BUGabundo> two from gmail using the most RAM
[21:49] <BUGabundo> 230 and 180MBs
[21:50] <BUGabundo> Private	 Proportional
[21:50] <BUGabundo> Chromium
[21:50] <BUGabundo> 1,277,864k	100,354k
[21:50] <BUGabundo> that's *a lot* of ram
[21:51] <fta> none of my 25 tabs uses more than 50M
[21:52] <BUGabundo> let me launch a brand new profile
[21:52] <BUGabundo> and reproduce same tabs
[21:52] <fta> open the ch task manager and keep it visible
[21:52] <fta> shift + esc
[21:54] <BUGabundo> I know
[21:54] <BUGabundo> I do
[21:55] <BUGabundo> flash is using 200MBs
[21:55] <BUGabundo> seesmic 76
[21:56] <BUGabundo> and all other under 50MBs
[21:57] <BUGabundo> Chromium 306,636k
[21:57] <BUGabundo> way less total ram
[21:58] <fta> that's with a --temp-profile?
[21:58] <BUGabundo> yes
[21:58] <BUGabundo> turning off most of my addons
[21:59] <BUGabundo> and restarting
[21:59] <BUGabundo> lets see if that makes a diff in mem usage
[22:00] <BUGabundo> fta: ahahaah its not freeing the process
[22:00] <BUGabundo> I started it in gdb and it aiting closign
[22:01] <BUGabundo> 1000      2979  0.0  0.0   4148   576 pts/2    S    21:31   0:00 /bin/sh /usr/bin/chromium-browser -g
[22:01] <BUGabundo> 1000      2982  1.3 13.0 779248 528844 pts/2   S    21:31   0:24 /usr/bin/gdb /usr/lib/chromium-browser/chromium-browser -x /tmp/chromiumargs.5X9aW5
[22:04] <BUGabundo> fta: ctrl+c aint working either
[22:04] <BUGabundo> :O
[22:04] <fta> hm, looks like the accelerated-compositing bug. thought it was disabled for now
[22:04] <fta> try --disable-accelerated-compositing
[22:04] <BUGabundo> kill aint working either
[22:04] <BUGabundo> sending -9
[22:04] <fta> is still have that in my /etc/chromium-browser/default
[22:04] <BUGabundo> that did it
[22:05] <BUGabundo> 104,052k	 on start
[22:05] <jlebar> Hey, guys.  Just a heads up: We're planning to switch the Mozilla nightly builds to gcc 4.5 tomorrow.
[22:06] <jlebar> We're seeing pretty substantial perf wins with gcc 4.5 plus -O3.
[22:07] <micahg> jlebar: thanks for the heads up, is this for Firefox only or comm-central as well?
[22:08] <jlebar> micahg, Just Firefox right now.
[22:08] <micahg> jlebar: ok
[22:08] <BUGabundo> [10427:10427:2149079486:ERROR:base/native_library_linux.cc(28)] dlopen failed when trying to open libGLESv2.so: libGLESv2.so: cannot open shared object file: No such file or directory
[22:08] <BUGabundo> [10342:10357:2221656145:ERROR:chrome/browser/extensions/extension_protocols.cc(134)] Denying load of chrome-extension://edacconmaakjimmfgnblocblbcdcpbko/favicon.ico from about:blank because the extension does not have access to the requesting page.
[22:08] <BUGabundo> fta: turning that flag on is messing one add on
[22:11] <fta> Session Buddy? that has nothing to do with accelerated hw
[22:13] <chrisccoulson_> jlebar, might want to take a look at bug 663294 ;)
[22:13] <ubot2> Launchpad bug 663294 in gcc-4.5 (Ubuntu) "Firefox built with gcc-4.5 is a non-starter on i386 with -pie (affects: 3) (heat: 182)" [High,Confirmed] https://launchpad.net/bugs/663294
[22:13] <chrisccoulson_> although, i guess that doesn't really affect your builds
[22:14] <micahg> chrisccoulson_: have you decided what you want to do about beta 7 re natty yet?
[22:14] <jlebar> chrisccoulson_, I believe we build with a patched gcc-4.5.  I'm not sure what the patch fixes, but maybe it's that.  :)
[22:15] <chrisccoulson_> micahg - upload as soon as all the major bits are ready (those being ubufox and translations)
[22:15] <micahg> chrisccoulson_: ok, should I go ahead and upload to the PPA this weekend then or should I wait for you to upload to Natty?
[22:16] <chrisccoulson_> jlebar, it's not the patch for this is it? https://bugzilla.mozilla.org/show_bug.cgi?id=594611
[22:16] <ubot2> Mozilla bug 594611 in js-ctypes "jsctypes busted on linux with gcc 4.5" [Normal,Resolved: fixed]
[22:16] <chrisccoulson_> oh
[22:16] <chrisccoulson_> wrong link
[22:16] <chrisccoulson_> d'oh
[22:16] <chrisccoulson_> jlebar, i meant http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45623 ;)
[22:16] <ubot2> gcc.gnu.org bug 45623 in tree-optimization "[4.5 Regression] GCC 4.5.[01] breaks our ffi on Linux64. ABI break?" [Normal,Resolved: fixed]
[22:17] <chrisccoulson_> micahg - we should probably wait first
[22:17] <jlebar> chrisccoulson_, that's probably it, yes.
[22:17] <micahg> chrisccoulson_: ok, any idea how long, I'll add a note to the /topic
[22:17] <BUGabundo> fta: what's "Signature verification failed" ?
[22:17] <micahg> release is tomorrow barring any respins
[22:17] <BUGabundo> when trying to install an addon?
[22:17] <jlebar> Well, I hope you guys get the -pie thing fixed so you can get this speedup.
[22:17] <jlebar> it's really quite nice.  :)
[22:18] <chrisccoulson_> jlebar, i just disabled -pie on our builds ;)
[22:18] <chrisccoulson_> but that's not a long term solution for us
[22:18] <fta> BUGabundo, either your addons is busted, or the api changed and you need to re-install it
[22:18] <BUGabundo> I'm doing it now
[22:18] <fta> -s
[22:18] <BUGabundo> and when I tried it showed that
[22:19] <jlebar> haha, I see.
[22:20]  * micahg wonders if hardening-wrapper needs a rebuild
[22:20] <BUGabundo> fta: reporting to lastpass dev team
[22:21] <BUGabundo> I guess soemthing changed
[22:24] <chrisccoulson_> micahg, i don't think so, it's just a perl script
[22:36] <micahg> chrisccoulson_: why is it arch specific then?
[22:38] <chrisccoulson_> micahg - not sure about that
[22:39] <BUGabundo> fta: fyi since I removed that app, and enabled --disable-accelerated-compositing it seems better
[22:39] <BUGabundo> could be a bad addon
[22:40] <micahg> chrisccoulson_: maybe for the test suite
[22:42] <fta> BUGabundo, most probably the flag
[22:42] <BUGabundo> [12531:12531:2876911258:ERROR:chrome/browser/extensions/extension_error_reporter.cc(55)] Extension error: Signature verification failed
[22:42] <BUGabundo> [12531:12949:2884052311:ERROR:net/base/x509_certificate_nss.cc(564)] No EV Policy Tag
[22:42] <BUGabundo> [12531:12531:2884503862:ERROR:chrome/browser/extensions/extension_error_reporter.cc(55)] Extension error: Signature verification failed
[22:42] <BUGabundo> [12531:12531:3009553900:ERROR:chrome/browser/extensions/extension_error_reporter.cc(55)] Extension error: Could not open crx file for reading
[22:43] <BUGabundo> seeing a few more in the logs
[22:43] <BUGabundo> I guess a lot changed in todays build