Kristian Rietveld (kris)
Tim Janik (rambokid)
Matthias Clasen (mclasen_)
Billy Biggs (vektor)   
Jody Goldberg (jody)  
Federico Mena Quintero (federico)
Behdad Esfabod (behdad)
Mart Raudsepp (leio)
Carol Spears (carol)

**** BEGIN LOGGING AT Wed Nov  9 21:11:44 2005

Nov 09 21:13:11 <kris>	we start in 45 minutes, right?
Nov 09 21:19:36 <rambokid>	that's my impression
Nov 09 21:47:12 -->	ebassi|dinner (~ebassi@213-140-6-105.ip.fastwebnet.it) has joined #gtk-devel
Nov 09 22:00:15 -->	johannes (~johannes@p5496A029.dip0.t-ipconnect.de) has joined #gtk-devel
Nov 09 22:00:56 <matthias_>	ok, 4pm here
Nov 09 22:01:41 <--	jean has quit (Leaving)
Nov 09 22:01:42 <kris>	ok
Nov 09 22:01:44 <matthias_>	should we start with the range arrow issue ?
Nov 09 22:02:18 <matthias_>	I looked through old meeting logs and old bugs, but couldn't find the discussion  I thought I remembered
Nov 09 22:03:02 <rambokid>	matthias_: well, i guess you could ask around if anyone else remembers...
Nov 09 22:03:23 <rambokid>	as far as i'm concerned, the funcitonality doesn't need to be configurable.
Nov 09 22:03:52 -->	behdad (~behdad@CPE000fb55e466d-CM014250035807.cpe.net.cable.rogers.com) has joined #gtk-devel
Nov 09 22:04:03 <matthias_>	does anyone else remember a discussion about making range arrows insensitive when the range is at an end ?
Nov 09 22:04:16 -->	richard (~rhult@ua-83-227-193-217.cust.bredbandsbolaget.se) has joined #gtk-devel
Nov 09 22:05:50 ---	ebassi|dinner is now known as ebassi
Nov 09 22:06:19 -->	jdahlin (~jdahlin@200-161-81-84.dsl.telesp.net.br) has joined #gtk-devel
Nov 09 22:06:40 <rambokid>	matthias_: do you remember anyone else who was involved?
Nov 09 22:06:51 <matthias_>	certainly owen
Nov 09 22:07:09 <matthias_>	but thats not really useful
Nov 09 22:07:19 <vektor>	Certainly would pose problems for anyone who overloads the meaning of those arrows.
Nov 09 22:07:45 <matthias_>	what would you overload them for ?
Nov 09 22:08:02 <rambokid>	what? i'd rather ask: how?
Nov 09 22:08:03 -->	federico (~federico@201.137.114.202) has joined #gtk-devel
Nov 09 22:08:12 <federico>	la la la
Nov 09 22:08:17 <rambokid>	vektor: you can't do that
Nov 09 22:08:19 <federico>	KRIS
Nov 09 22:08:26 <federico>	kris: I need to reply to your lovely lovely mail
Nov 09 22:08:35 <federico>	kris: it's mostly sane
Nov 09 22:08:38 <federico>	good job!
Nov 09 22:08:39 <vektor>	Editors that expand as you scroll to the right.
Nov 09 22:08:44 <kris>	federico: great
Nov 09 22:09:13 <matthias_>	ok, lets keep the sensitivity toggling in for now. If we come up with a compelling reason, we can add a setting to turn it off
Nov 09 22:09:27 <rambokid>	vektor: that is unrelated. the arrow sensitivity just reflects the slider position.
Nov 09 22:09:33 <rambokid>	matthias_: ok.
Nov 09 22:09:57 <federico>	is this about scrollbars desensitizing the arrows when the slider bumps into them?
Nov 09 22:10:07 <federico>	gnumeric will hate that
Nov 09 22:10:22 <vektor>	rambokid: isn't it the same thing, you need that button to still be clickable when the thumb is at the end so that the application can expand the size of the editor?
Nov 09 22:11:06 <matthias_>	federico: yes. why will gnumeric hate it ?
Nov 09 22:11:44 -->	jody (~jody@72.59.8.128) has joined #gtk-devel
Nov 09 22:11:50 <federico>	matthias_: the scrollbar's area is the nonempty area of the spreadsheet.  But you can click on the arrows to scroll past that.
Nov 09 22:12:02 <federico>	matthias_: i.e. the scrollbars don't map to A0..IV65535
Nov 09 22:12:30 <matthias_>	isn't that roughly the use case vektor was describing ?
Nov 09 22:12:41 <behdad>	it is
Nov 09 22:12:54 <federico>	matthias_: sorry, I got in late :)
Nov 09 22:13:01 <federico>	but yes
Nov 09 22:13:24 <matthias_>	hmm, so this sounds more like we need this to be switchable per-application, rather than a setting
Nov 09 22:13:36 <rambokid>	federico: you can click on the arrow and cause an action when the trough is at the end?
Nov 09 22:13:59 <rambokid>	matthias_: per widget actually.
Nov 09 22:14:06 <matthias_>	right
Nov 09 22:14:37 <federico>	rambokid: yes
Nov 09 22:14:38 <federico>	woah
Nov 09 22:14:41 *	federico tries it
Nov 09 22:14:47 <federico>	rambokid: actually, it's a lot more magic than that
Nov 09 22:14:49 <matthias_>	what happens ?
Nov 09 22:14:53 <federico>	jody: that's some *great* behavior
Nov 09 22:14:56 <federico>	run gnumeric
Nov 09 22:15:01 <federico>	write something in A1
Nov 09 22:15:05 <federico>	write something in A50
Nov 09 22:15:08 <federico>	scroll to A1
Nov 09 22:15:14 <federico>	now play with the slider and the arrows
Nov 09 22:15:26 *	matthias_ installs gnumeric
Nov 09 22:15:32 *	behdad too
Nov 09 22:16:09 <federico>	how can you live without gnumeric?
Nov 09 22:16:17 <behdad>	federico: with bc :-P
Nov 09 22:16:27 <matthias_>	I am not a manager...
Nov 09 22:16:30 <behdad>	and join, and tr.
Nov 09 22:16:30 <vektor>	I've never tried it myself, really.
Nov 09 22:16:33 <federico>	all my expense reports go through gnumeric
Nov 09 22:16:43 <behdad>	federico: try gnuchash
Nov 09 22:16:50 <behdad>	s/ch/c/
Nov 09 22:16:56 <behdad>	and you may port it to gtk2
Nov 09 22:16:58 <federico>	behdad: novell does expense reports with a stupid pre-made excel spreadsheet
Nov 09 22:18:07 <matthias_>	ok, I'm going to reopen the bug about range arrow sensitivity
Nov 09 22:19:26 <behdad>	wow
Nov 09 22:19:29 <matthias_>	argh
Nov 09 22:19:29 <behdad>	so nice
Nov 09 22:19:56 *	matthias_ installed gnumeric on his work machine instead of locally..
Nov 09 22:21:15 <matthias_>	should we talk about async filechooser api changes ?
Nov 09 22:23:20 <matthias_>	the gnumeric behaviour seems to advocate toggling sensitivity for each arrow separately
Nov 09 22:23:33 <matthias_>	since it does something on the right, but not on the left...
Nov 09 22:24:02 <rambokid>	matthias_: and per widget, since the rest of the gnumeric dialogs shouldn't change
Nov 09 22:24:02 <jody>	behdad: they're almost finished porting gnucash to gtk2.  They're using libgoffice for the charting
Nov 09 22:24:11 <behdad>	jody: nice
Nov 09 22:24:38 <behdad>	jody: btw, I cannot scroll using the scroll bar when the page is empty, should start with the arrow, then the scroll bar works
Nov 09 22:25:24 <matthias_>	rambokid: thats what I meant
Nov 09 22:25:25 <rambokid>	matthias_: can we discuss the FC async API per email? i need more time to answer kris' proposa and i don't think we'll finish that in 30 minutes
Nov 09 22:25:51 <matthias_>	sure
Nov 09 22:25:58 <behdad>	matthias_: maybe have a method to change sensitivity on one of the arrows, and the default handler that calls them.  gnumeric can subclass and override that
Nov 09 22:26:13 <rambokid>	matthias_: well, you said per-arrow, and i think you had a point there, i.e. we probably really need to distinguish bottom/top arrows in the setting
Nov 09 22:26:54 <rambokid>	behdad: i think a simple setter/getter API for a per-arrow flag will be easier
Nov 09 22:27:30 <behdad>	rambokid: how can gnumeric use it then?
Nov 09 22:27:44 <behdad>	note that it's not like in gnumeric the right arrow is always sensitive
Nov 09 22:27:51 <behdad>	it does have an end, which is IV
Nov 09 22:28:12 <behdad>	and for the bottom arrow, it's 65536
Nov 09 22:28:33 <matthias_>	maybe jody can outline what he thinks he would need
Nov 09 22:29:11 <federico>	matthias_: kris's plan is very good; we already talked about it over mail
Nov 09 22:29:19 <rambokid>	matthias_: tko isn't here... anyway he already bloged about it: http://blogs.gnome.org/view/tko/2005/11/06/0
Nov 09 22:29:25 <federico>	matthias_: I'll just answer to his mail to gtk-devel-list
Nov 09 22:29:51 <matthias_>	I'll answer too
Nov 09 22:29:58 <federico>	the one thing that's left is how to make older/newer gnomes don't crash.  I do like the idea of using $(prefix)/whatever/2.6.0/filesystems
Nov 09 22:29:59 <jody>	behdad: good point.  It seems to get disabled when the scrollbar touches the top
Nov 09 22:30:18 <rambokid>	matthias_: so we're wokring on backfolding maemo-gtk changes into upstream where it makes sense, and this is where the status is kept: http://maemo.org/maemowiki/MaemoGtk26Contributions
Nov 09 22:30:29 <matthias_>	federico: thats the right approach, imo
Nov 09 22:30:41 <matthias_>	rambokid: thanks for the pointer
Nov 09 22:30:43 <rambokid>	behdad: the same way it works currently. basically the flag would just toggle the newly added behaviour. 
Nov 09 22:30:51 <matthias_>	maybe somebody should blog that, for public excitement...
Nov 09 22:31:39 <behdad>	rambokid: the proposal is to add a setter/getter to desensitize the arrow OR to turn on/off automatic desensitization of the arrows?
Nov 09 22:31:46 <kris>	i need to blog about 2.10 again, indeed ;)
Nov 09 22:32:03 <kris>	also, i think it's also a good idea to discuss FC on the list first, and maybe do the leftovers in next week's meeting
Nov 09 22:32:35 <rambokid>	behdad: no proposal just brainstorming. if you really want to investigate and comment the issue, please do that in the bug report, lemme dig up the number for you
Nov 09 22:32:54 <rambokid>	http://bugzilla.gnome.org/show_bug.cgi?id=321056
Nov 09 22:32:59 <behdad>	thanks
Nov 09 22:33:52 <federico>	what's the status on printing?
Nov 09 22:33:55 <jody>	matthias_: after a quick look at the gnumeric code it looks like we use the 'adjust-bounds' signal to massage things
Nov 09 22:34:29 <jody>	matthias_: so my initial response would be to disable the de-sensitization of the arrows if there is an adjustbounds handler
Nov 09 22:34:49 <jody>	but that is not terribly elegant.
Nov 09 22:34:57 <rambokid>	why does gtk head still use lib/gtk-2.0/2.4.0/ and not 2.10 ?
Nov 09 22:35:19 <matthias_>	jody: maybe things can get arranged so that the insensitivation doesn't happen if the adjust-bounds handler adjusts the bounds...
Nov 09 22:35:20 <rambokid>	(that was re $prefix as raised by federico)
Nov 09 22:35:41 <behdad>	jody: but it would be cool if you could still desensitize it when at the real real end.
Nov 09 22:35:43 <matthias_>	rambokid: because thats ABI version
Nov 09 22:36:13 <matthias_>	behdad: that would work automatically if the desensitivation would play nicely with adjust-bounds
Nov 09 22:36:50 <behdad>	matthias_: yes, that sounds good
Nov 09 22:36:50 <jody>	behdad: agreed.  It would be better if adjust-bounds could set the sensitivity
Nov 09 22:36:55 <federico>	behdad: you have mail
Nov 09 22:37:04 <jody>	actually.  it probably can
Nov 09 22:37:54 <behdad>	matthias_: not sure that it's possible though.  when the slider is at the end, the bounds are not adjusted before the user tries to scroll again, right?
Nov 09 22:38:50 <behdad>	federico: thanks a lot.
Nov 09 22:39:19 <matthias_>	behdad: I envisioned automatic desensitiviation would be done if the value is still at the end after the adjust-bounds emission 
Nov 09 22:39:37 <federico>	behdad: and you are lucky; you got some pretty colors
Nov 09 22:39:52 <behdad>	federico: any idea about that te language?
Nov 09 22:40:06 <behdad>	federico: weird.  something's not right.  I'll figure out
Nov 09 22:40:56 <federico>	behdad: let me re-run the benchmark just for that one
Nov 09 22:41:07 <federico>	holy shit, I love my benchmark
Nov 09 22:41:34 <federico>	fuck, it took even longer
Nov 09 22:41:59 <behdad>	federico: get the benchmark framework out, make it load .so test modules. :-D
Nov 09 22:42:25 *	federico re-runs with $many iterations
Nov 09 22:42:32 <rambokid>	matthias_: oh and a quick note on the gmem.c changes. the API is supposedly there, but we still have no sliced bread/memory implementation ;) 
Nov 09 22:42:47 <matthias_>	rambokid: I noticed that
Nov 09 22:43:19 <rambokid>	matthias_: it'll take me longer than originally envisioned, after reading the memory part of your patch i read up on magazines in the old and new slab papers and then some more.
Nov 09 22:43:59 <matthias_>	its not that urgent
Nov 09 22:44:02 <rambokid>	matthias_: at this point, i have a pretty good idea of what we should have (basically most of the logic of the 2001 paper). 
Nov 09 22:44:22 <matthias_>	although...
Nov 09 22:44:42 <matthias_>	people have been asking if all of the recent performance work will be in gnome 2.14
Nov 09 22:44:59 <rambokid>	matthias_: to shorten that: it should have magazines but also support freeing and do cache colorization to avoid cache line aliasing. 
Nov 09 22:45:35 <rambokid>	matthias_: it'll land in the next couple weeks, just can't tell you the exact day.
Nov 09 22:45:39 <matthias_>	rambokid: I trust you to come up with something at least as good as my prototype...
Nov 09 22:45:50 <matthias_>	err, no, trust is a bad word :-)
Nov 09 22:45:54 <rambokid>	heh ;)
Nov 09 22:46:08 <behdad>	let's get to the release schedules
Nov 09 22:46:51 <carol>	matthias_: i think it is okay for you to trust rambokid. he just does not trust you...
Nov 09 22:48:13 <matthias_>	I don't know owens plan for pango, or in fact, if he has one
Nov 09 22:48:45 <federico>	behdad: "te" keeps on sucking
Nov 09 22:48:54 <behdad>	federico: weird.  will look into.
Nov 09 22:49:10 <matthias_>	but I don't think backporting all the performance work to the stable branch is a good investment
Nov 09 22:49:28 <federico>	matthias_: it would be nice to have a generally-faster gnome 2.14
Nov 09 22:49:34 <behdad>	matthias_: last time we talked about it, we decided to go with gtk+ schedule
Nov 09 22:49:46 <behdad>	matthias_: but there was nothing like these performance work planned
Nov 09 22:50:45 -->	gmt (~gentoo@host103-25.pool8251.interbusiness.it) has joined #gtk-devel
Nov 09 22:50:47 <behdad>	the backporting is really a matter of applying the diffs, but performance work on the stable branch doesn't sound right to me.
Nov 09 22:51:02 <matthias_>	pango head depends on glib head, so if we decide to do a short cycle for pango, we need to do one for glib too
Nov 09 22:51:48 <behdad>	matthias_: if need be, we can push back that linebreaking patch
Nov 09 22:51:50 <matthias_>	which is fine for me, once rambokid lands g_slice
Nov 09 22:52:16 <behdad>	yeah, sounds good
Nov 09 22:53:38 <rambokid>	matthias_: what do you mean with that? release glob-2.10 soon but gtk+-2.10 later?
Nov 09 22:53:59 <rambokid>	glib not glob of course ;)
Nov 09 22:54:02 <matthias_>	rambokid: yes
Nov 09 22:54:19 <rambokid>	matthias_: rough ideas on dates?
Nov 09 22:54:53 <leio>	Would then gtk+-2.10 release come together with a glib-2.12, bringing the numbers out of sync?
Nov 09 22:55:01 <matthias_>	rambokid: gnome 2.14 is due in february, I think
Nov 09 22:55:35 <matthias_>	rambokid: if we decided to go this route, we should probably do 2.9 releases soon after you land the g_slice implementation
Nov 09 22:56:52 <rambokid>	matthias_: there's a problem with that btw. gtktree*something was doing invalid GAllocator business and broke after i removed the GAllocator implementation. so i had to fix gtk HEAD on this.
Nov 09 22:57:14 <rambokid>	matthias_: for 2.8 that means this fix needs to go into that version as well to work with glib head.
Nov 09 22:57:17 <matthias_>	rambokid: oh, I think I worked around it in glib head...
Nov 09 22:57:44 <matthias_>	by setting n_allocs to a nonzero value in the dummy GAllocator
Nov 09 22:58:14 <rambokid>	matthias_: i hacked the GAllocator use out of head. i'm talking about that code still being in gtk-2-8, so it'd have problems with glib-head
Nov 09 22:58:36 <matthias_>	rambokid: let me find the bug 
Nov 09 23:00:23 <behdad>	federico: just to make sure, you got the ltr-rtl patch, not the ltr-only, right?
Nov 09 23:01:11 <jody>	matthias_: is there a bug I can reply on re: arrow sensitization ?
Nov 09 23:01:12 <matthias_>	rambokid: http://bugzilla.gnome.org/show_bug.cgi?id=320755
Nov 09 23:01:39 <matthias_>	jody: http://bugzilla.gnome.org/show_bug.cgi?id=321056
Nov 09 23:01:40 <behdad>	jody: http://bugzilla.gnome.org/show_bug.cgi?id=321056
Nov 09 23:03:24 <matthias_>	ok, do you think it is reasonable to try to shorten the cycles for glib and pango ?
Nov 09 23:04:03 <federico>	behdad: I got the last patch in the bug
Nov 09 23:04:14 <behdad>	matthias_: I'm with.
Nov 09 23:04:27 <federico>	behdad: http://bugzilla.gnome.org/attachment.cgi?id=54511&action=view
Nov 09 23:04:30 <behdad>	federico: ok thanks.  doesn't make any sense though that rtl didn't get better
Nov 09 23:04:37 <matthias_>	anyway, need to go, family is waiting for dinner. See you next week
Nov 09 23:04:50 -->	jean (~jean@che21-1-82-239-125-56.fbx.proxad.net) has joined #gtk-devel
Nov 09 23:04:51 <federico>	behdad: maybe there's a lot of bidi strings?
Nov 09 23:05:07 <behdad>	federico: figuring out.
Nov 09 23:05:11 <--	matthias_ has quit (Remote closed the connection)
Nov 09 23:05:17 <behdad>	federico: it's not about lots, it's about all.
Nov 09 23:06:29 <--	jdahlin has quit (Ex-Chat)
Nov 09 23:06:30 <federico>	behdad: uh oh
Nov 09 23:08:41 <federico>	behdad: what's throwing them off?  the .dat files will have a lot of "%s" shit in them
Nov 09 23:08:46 <--	johannes has quit (... und tschüß)
Nov 09 23:08:54 <federico>	arabicarabicarabic %s arabicarabic
Nov 09 23:09:06 <behdad>	federico: right.  maybe.
Nov 09 23:09:19 <behdad>	not quite all, but yes, a lot
Nov 09 23:09:26 <behdad>	lemme gather some stats
Nov 09 23:09:58 <behdad>	federico: I got the short circuit path hit when typing persian in gedit. 
Nov 09 23:10:18 <behdad>	it's not quite useless even if most messages don't get into it ;)
Nov 09 23:11:53 <leio>	wouldn't in real-life usage the %s be already replaced with rtl strings at the point pango gets it?
Nov 09 23:12:28 <--	_Legion_ has quit (Ex-Chat)
Nov 09 23:13:12 <behdad>	leio: well, most of the time file names, etc get that place
Nov 09 23:13:26 <leio>	hmm, true
Nov 09 23:13:51 <--	jean has quit (Leaving)
Nov 09 23:13:52 <leio>	also numbers. those were ltr always and make it bidi?
Nov 09 23:14:10 <behdad>	leio: yes.
Nov 09 23:14:29 <rambokid>	kris: you're blogging about this, right?
Nov 09 23:14:49 <leio>	oh well. At least some labels would speed up, and you're dealing with getting better profiling results for that already, I understood :)
Nov 09 23:15:24 <kris>	rambokid: when i get time agian
Nov 09 23:15:27 <kris>	rambokid: maybe this weekend
Nov 09 23:17:55 <rambokid>	good
Nov 09 23:18:25 <federico>	hmm
Nov 09 23:18:37 <federico>	maybe pango-language-profile should scan the strings when loading the .dat files
Nov 09 23:18:40 <behdad>	did we decide about the schedule?
Nov 09 23:18:45 <federico>	and replace percent-thingies by something reasonable
Nov 09 23:18:48 <federico>	%s by spaces
Nov 09 23:18:54 <federico>	%d by 123
Nov 09 23:19:02 <behdad>	yeah, sounds better
Nov 09 23:19:11 <behdad>	still spaces are not necessarily realistic
Nov 09 23:19:11 *	federico eyes behdad
Nov 09 23:19:59 <federico>	behdad: well, pango-language-profile will have a hard time picking a suitable string... I assume that if you substitute a string inside another one, then that string will also have the same probability of being bidi as the whole language
Nov 09 23:20:08 <federico>	now
Nov 09 23:20:13 <federico>	note that I flunked my statistics class
Nov 09 23:20:18 <behdad>	federico: no, I say go for abcd
Nov 09 23:20:19 <federico>	so I'm talking out of my ass
Nov 09 23:20:40 <behdad>	most of the time it's something in latin
Nov 09 23:21:17 ---	You have left channel #gtk-devel
**** ENDING LOGGING AT Wed Nov  9 23:21:17 2005