Meeting on irc.gnome.org:#gtk-devel Meeting started September 20 2004 17:05 EST (21:05 UTC) In attendence: Anders Carlsson (andersca), Matthias Clasen (matthias), Owen Taylor (owen), Jonathan Blandford (jrb), Billy Biggs (vektor), Robert Ă–gren (roboros), John Ehresman (jpe), Soeren Sandmann (ssp), Mark McLoughlin (markmc), Carlos Garnacho (garnacho), Jody Goldberg (jody) it's time!! once again I failed to mail out any agenda items but I am proud to welcome markmc of release team fame today hrrm dubious honour that :) so, what do we want to discuss today ? uh how's 2.6 looking? matthias: Well, we might as well start off with markmc's item, which was status of gtk+-2.6 for gnome-2.10 ok, lets do that first yeah, so basically all I'm wondering is can we let GNOME developers start using the new APIs so, looking at the feature list on www.gtk.org/plan/2.6, we don't look too bad, missing only the filechooser enhancements which federico wanted to do are ye still hoping to freeze soon etc. markmc: We (or well, not me, but Matthias, etc) seem to be chugging along pretty well; all the bug stuff is in and just needs testing in practice markmc: I haven't heard any other opinions, so I assume we'll go into a slushy freeze around mid-October matthias, great, there'd be no harm in starting to test the APIs now then matthias, we'd be looking at roughly january for API freeze in GNOME I guess the new pieces certainly have some holes and bugs, but those should be getting addressed before too long... yeah markmc: that fits nicely with our 2.6 target date of mid-December super * markmc looks forward to killing eel-ellipsizing-label :) matthias, do we have a list of new APIs generated from Since: 2.6 ? markmc: having people try the new stuff soon will certainly help finding those bugs... markmc: I hope that the api docs shipped with 2.5.3 contain separate indices for "new since x.y" stuff matthias, ah, awesome matthias, I'll look at that, thanks that's me happy anyway :) markmc: what we're currently missing is chapters for the migration part detailing how to port from libgnomeui widgets to their new gtk counterparts markmc: I hope that I can get started on that for the next 2.5.x release in ~2 weeks matthias, just the list of APIs will at least make people aware of the new stuff anyway what would be the chances of http://bugzilla.gnome.org/show_bug.cgi?id=139486 going into 2.8, provided there's at least working prototype by time 2.6 ships? markmc: just checked. there are in fact separate indices for 2.2, 2.4, 2.6 and deprecated symbols matthias, rocking mathrick: I'd say relatively good. mathrick: I would say that the chances are good, but I think before prototyping, you should maybe discuss your design on gtk-devel-list matthias: yeah, currently we're tossing around ideas with ed__ and jamesh, when it's some more concrete it'll go to -devel mathrick: sounds good cool btw, meetings are done weekly at this time? mathrick: unless a holiday or some other circumstance gets in the way. I usually announce the meetings to gtk-devel-list sometime between Friday and Monday morning... matthias: yeah, I see announces every now and then, just wanted to get general frequency :) so, speaking about releases...I have done some over the weekend. we decided to do 2.4.10 earlier to get the gdk-pixbuf fixes out hmm, one more thing, is this channel persistent, or only exists at time of meeting and normal devel discussion goes into #gtk+? but I think we should perhaps still target the first week of October for another 2.4.x release... mathrick: I think the channel exists as long as people forget to leave it... mathrick: Normal devel discussion goes into #gtk+, there's usually one or two people who don't bother to leave this channel, but it's not used fair enough markmc: while you are here: when is gnome 2.8.1 expected ? matthias, the plan last time around was .1 a month after .0, .2 two months after that and .3 three months after that matthias, so in a few weeks time probably matthias, but we haven't drawn up a schedule yet markmc: ok, so targetting first week of October for 2.4.11 should be in time for 2.8.1... matthias, yeah, that'd be sweet there are still plenty of file chooser quirks to work out til them... matthias: the HIG dialog API never landed, right? jrb: not yet. I think garnacho is still working on a revised patch... okay garnacho: are you ? jrb: good point, though. Maybe everybody should look over his list of favourite api additions which never quite made it before the api freeze so that we don't drop too many useful additions just because nobody looked at them in time... matthias: sure, I needed some time after the exams, but I'm already working on it garnacho: good to know, thanks np :) one thing I wanted to point out regarding the 2.5.3 releases is that I added the plt-reduction-by-aliasing to glib and gobject last week and again found a few dark corners of the abi which needed cleaning up, like forgetting to stip _-prefixed symbols, or accidentally exported symbols which are in no headers, etc so if you find any problems due to symbols which are no longer there, complain... the good aspect of this cleanup work is that we have a clearly defined (and checked) abi now, and the .def is derived from it ssp: do you know of problems with the redocking the toolbar in a different orientation ? Is that entirely a bonoboui problem ? matthias: You mean changing from horizontal to vertical? I don't know of any problems wrt. to that. ssp: yes, if you do that in gedit, the toolbar looses all icons, and a ton of warnings are spewed matthias: In my version of GEdit, you can't change the orientation of the toolbar matthias: 2.7.92 matthias: That version also uses the BonoboToolbar ssp: you can detach it, then move it over the left edge of the window to dock it there, no ? federico did some fixes to BonoboToolbar at some point; might be worth tracking him down matthias: Ah, right. I turned that off a long time ago - let me try matthias: Yeah, I see the problem, but GEdit uses BonoboToolbar, so I don't think I can be blamed :-) ssp: I wasn't trying to I'll bug federico about it... matthias: I hope to get the arrow toolbutton in 2.6 ssp: that would be nice hey jody matthias: I've missed the meeting haven't I ? jody: It's definitely on it's last legs maybe you can revive it... I'd like to make a few extensions to GtkUIManager At least one of which would involve some semantic changes 1) Is there time to get something into 2.6 if I get a proposal out in the next two weeks 2) Is it reasonably to even consider semantic changes ? owen, any chance of moving forward on the win32 grab bug? jody: By semantic changes you mean "incompatible semantic changes"? jpe: Not my area. jody: 1) we're trying to go to slush freeze around the middle of October, 2) depends on the changes, I'd say... owen: yes. I'd like to see all actions with accelerators automatically have accelerators, not just those listed in menus owen, whose area then? jody: Hmm. I think that would be very confusing; we decided if I'm not mistaken against that explicitely jpe: win32 is all tml jpe: Or you are asking about the grab-broken signal? jpe: For that, would be best to mail gtk-devel-list, I'll try to follow up there as appropriate owen, yes, the grab-broken signal owen: I'll make up a proposal to explain why I see that, and a few other things as necessary jody: I don't think introducing visibilty of hidden actions to the user would be sufficiently compatible .. I could see it breaking the heck out of an application with, say, two modes owen, ok jody: There might be some way of compatibility allowing apps to enable it if desired jody: I think there is prior discussion on this issue around somewher,e but I'm not sure how to find it. owen: that's what I suspected. Although to me the current approach seems like a recipe for missing actions owen: modes seem well handled by sensitivity rather than visibility jody: One of jamesh, mclasen, or ssp might remember more on the issue jody: good to write it up and send to gtk-devel-list in any case, owen: i would have to dig through the archives first (/me was missing mnemonics for toolbars when doing http://fishsoup.net/blog_images/Hinting-Viewer-1.png...) owen: I think the rationale is that a toolbar is supposed to be a shortcut for mouse users, so mnemonics would be clutter ssp: Please expand on 'clutter' they're not displayed. ssp: Yeah, I know, I wasn't really saying they shoudl be added, just that I missed them. jody: Talking about my comment, not yours owen: That's exactly the sort of problem I've seen a few times. We add an accelerator for something, but it's in a toolbar not a menu and the accel is dropped until we manually list it jody: since the ui descriptions are xml, it should be possible to extend the format and add something for toolitems with accelerators without breaking compat... jody: Well, my actions were in the toolbar *and* the menu; I just wanted obvious keyboard shortcuts to them. matthias: possibly, most of my proposed additions are in a similar vein matthias: eg the ability to override a label in the xml, rather than in the action or to define an additional accelerator in an accelerator node that sort of thing In this instance I'd prefer some sort of GtkUIManager hook jody: will you write a mail to gtk-devel-list ? GtkUIManager::only_menus_get_acclerators_by_default matthias: I'll send something on friday. I just wanted to check to see if it would fit with the schedule * matthias has leave now; see you later aaaaargh federico: a little late... federico: have you touched BonoboToolbar? federico: don't worry, you were blamed when absent ;) jrb: a long time ago jrb: I was trying to fix their lifecycle mess relatively recently --- when things got moved to GtkToolItem, etc., but didn't finish jrb: but that's in a branch Meeting ended September 20, 18:15 EST (22:15 UTC)