Meeting on irc.gnome.org:#gtk-devel Meeting started August 9 2005 16:02 EDT In attendence: Jonathan Blandford (jrb) Morten Welinder (gmorten) Matthias Clasen (mclasen) Tor Lillqvist (tml) Kristian Rietveld (kris) Owen Taylor (owen) Sebastien Bacher (seb128) Jody Goldberg (jody) Federico Mena Quintero (federico) tml: do you see the crash from 304702? (the assert, not necessarily trigger that exact way.) ok, so lets start then let me give a quick update on the current state of things owen and I discussed the schedule for 2.8 a bit today, and we think it would be good to try and get 2.8 releases out around the end of this week which should allow us to depend on cairo 0.9.1 and avoid some library dependency problems with libpixman do we know when cairo 1.0 is going to be released? in time for gnome 2.12 according to the plans kris: plan is before the GNOME RC freeze - http://live.gnome.org/ReleasePlanning_2fTwoPointEleven ok I don't know if rambokid can land atomic toggle and weak references this week but we think it will be ok to land them in 2.8.x, x > 0 the whole atomic reference counting needs some evaluation from the gstreamer camp anyway, when wim comes back gmorten: no ;-) I would like to know if people have any input on release notes/announcement items for 2.8 my current thinking can be found at people.redhat.com/mclasen/gtk-2.8.0-announcement.txt people.redhat.com/mclasen/gtk-2.8.0-notes.txt mclasen: is there any bug you have planned to fix before 2.8.0 ? one thing that should be mentioned is that the giochannel code for sockets on win32 has been rewritten, and any non-trivial use cases people have need to be verified that they still work hrm, am i not part of the team anymore? seb128: no, I don't have any bugs on the radar right now k tml: ok, thanks. Any other relevant win32-specific changes I missed ? I'm quite bothered with this icon scale one, but that's probably a GNOME 2.12 rather than a GTK 2.8 ... and since you guys don't have the issue anyway mclasen: maybe the release notes should also mention the note about reusing cell renderers seb128: I thought federico reproduced that one, didn't he ? kris: good point mclasen: perhaps you could mention that the GdkEventBroken happens much more easily and regularily on windows mclasen: no clue, owen said on the bug he doesn't get it so somebody else will have to debug, and you said you don't get it neither tml: any chance of you proxy debugging the issue with one of the 42 different reporters who haven't been able to themselves? seb128: speaking of changes we are still contemplating for 2.8, owen brought up the issue of libXft seb128: we're currently linking against it, but are not using it anymore what about libXft? oh gmorten: well, the solution is usually simple: don't try to use random odd fonts. and as owen says, that error can really be causd by many widely different things tml: Is it still occuring for random odd fonts? gmortne: DOes gnumeric have a drop down font selector where fonts show up "in themselves"? owen: well, it haven't occured for me in a long time I really don't think an assert is appropriate just because we don't like a given font. Messing up the display, perhaps, but not crash. tml: The issue with the system font being a bitmap font and not supporting bitmap fonts is fixed? owen: not yet. owen: someone is working on it. ivan wong is working on it, i think? mclasen: vektor just pointed out to me something on #cairo that makes the Xft issue not a ABI issue owen: nice (Though conceivably a minor API issue, but I'm much less worried about that, since using Xft but not linking to it is definitely a bug) crap, here I am missing the meeting again :) $ pkg-config --libs gtk+-2.0 -L/opt/rendering/lib -L/opt/rendering//lib -lgtk-x11-2.0 -lgdk-x11-2.0 -latk-1.0 -lgdk_pixbuf-2.0 -lm -lpangoxft-1.0 -lpangocairo-1.0 -lpangox-1.0 -lpangoft2-1.0 -lfreetype -lz -lcairo -lpango-1.0 -lgobject-2.0 -lgmodule-2.0 -lglib-2.0 -lfontconfig -lpixman -lXrender -lpng12 -lX11 -ldl ah yes, i just noticed his recent comment to #307469 owen: we do not, it was a performance nightmare jody: Not suprised mclasen: the -lpangox and libpangoxft ther eare actually useful. is working on a potentially interesting approach that would cache the previews as pixmaps rather than loading every font in existence every time http://www.abisource.org/~rob/tmp/go-combo-font.tgz I can forward the latest copy if you're interested tml: Of course, I do think we actually should be honoring the system default setting even if it is a bitmap font... I'm going on vacation hopefully tomorrow or the day after for two weeks right now I'm stuck with a gnome-vfs bug for work federico: congrats But displaying some other font would be certainly better than nothing owen: you mean "even if it is a truetype font"? yeah, isn't the "system default" supposed to be used only for title bars or something like that? (on the vacation, not the gnome-vfs bugs...) so I can't look into seb's bug about large icons, or the libgnomeui/file-chooser crasher mclasen: thanks; we'll move to our new house during that period :) tml: Isn't the default that the ms-windows theme was setting the setting for dialog box text, etc? federico: you might be interested to hear that I'll probably end up doing a 2.6.10 release, after all federico: to fix some grab breakage I introduced in 2.6.9 <-- GTKool_2kx has quit (Download Gaim: http://gaim.sourceforge.net/) owen: i'm not sure right now. i think at least earlier it used some api that if you read the docs carefully returned a font that is to be used in title bars and common controls menus mclasen: oh, interesting tml: Well, I'll leave it to the people who are more knowledgable :-) mclasen: even after the last 2.6 release, I may commit stuff to that branch occasionally (if bugs come up at work) hmm if during the 2.8 timeframe I find performance hotspots, can the fixes go into 2.8, or should they wait until 2.9? owen: hmm, modules/engines/ms-windows/msw_style.c:get_system_font() uses SystemParametersInfo(SPI_GETNONCLIENTMETRIC) to get varios fonts. the "NONCLIENT" sounds to me like it means windoe decorations and system menu. not something gtk should use federico: depends on the kind of fix, I would say tml: Hmm, presumably there is some way of getting the current dialog font? mclasen: I'm expecting things like "make resizing faster by having gtklabel not re-measure its layout if it is not insensitive, not ellipsized, etc." kris: you're back... ok, I have to leave a bit early today see you all next week Meeting ended 16:49 EDT