i18n
LOGS
05:02:26 <tagoh> #startmeeting i18n
05:02:26 <zodbot> Meeting started Thu Feb  3 05:02:26 2011 UTC.  The chair is tagoh. Information about MeetBot at http://wiki.debian.org/MeetBot.
05:02:26 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
05:02:32 <tagoh> #meetingname i18n
05:02:32 <zodbot> The meeting name has been set to 'i18n'
05:02:45 <tagoh> #topic agenda
05:02:54 <tagoh> https://fedoraproject.org/wiki/I18N/Meetings/2011-02-03
05:03:16 <tagoh> who else here today?
05:04:38 <paragan> hi
05:05:02 <juhp> hi
05:05:52 <dueno> hi
05:06:10 <tagoh> hi
05:06:53 <tagoh> ok, let's get started.
05:06:59 <tagoh> #topic F15
05:07:35 <dingyichen> hi
05:07:58 <tagoh> Feature Freeze and branching f15 from rawhide is coming next Tuesday, 8th of February.
05:09:00 <tagoh> so if you are planning to make any changes of deps, adding any features etc, please make sure if you've done that before that
05:09:50 <tagoh> guess we may miss new fontconfig and fixes in the font packages against it perhaps, but anyway
05:10:21 <juhp> yay to branching
05:11:12 <tagoh> any updates on features?
05:12:21 <juhp> I reviewed the text and improved it a both for both
05:12:52 <juhp> I think paragan said there were some more questions from fesco (for typingbooster at least)
05:12:52 <tagoh> cool
05:13:13 * juhp is not sure he will ever grow to like that name... ;)
05:13:27 <tagoh> sure. from the meeting log, it seems approved though. good to clarify if there are any questions :)
05:14:16 <paragan> yes they were asking about forking of ibus-table
05:14:40 <juhp> ah
05:14:45 <tagoh> aha
05:14:56 <juhp> I thought we should discuss it here....
05:14:59 <paragan> and whether indic ibus-table and ibus-table can be installed without any conflict
05:15:02 <juhp> maybe in the IM section?
05:15:10 <tagoh> juhp: ok.
05:15:14 <juhp> sure it would be renamed...
05:15:40 <juhp> but it is a F15 feature - so maybe makes more sense to take it here dunno
05:15:52 <tagoh> I don't mind either
05:15:58 <juhp> pravins: around?
05:16:59 <pravins> juhp: yeah
05:17:02 <tagoh> btw no issues to proceed RupeeSign feature so far?
05:17:07 <pravins> hi all
05:17:30 <tagoh> pravins: hi
05:18:11 <paragan> tagoh, yes RupeeSign feature is approved and ok now
05:18:22 <tagoh> paragan: ok, good
05:18:40 <pravins> I will check log once and answer those questions
05:19:02 <juhp> paragan: I think we already answered the question about forking - was there a newer question?
05:19:44 <pravins> yeah, right
05:20:10 <paragan> juhp, no new questions then
05:21:18 <tagoh> any concerns or something you want to raise here for f15 so far?
05:21:41 <juhp> pravins: although I have kind of said might be ok to fork if we have to - I would prefer to fix ibus-table first and then see if really need to fork it now or later
05:22:27 <pravins> juhp: ibus-table need support of wildcard
05:22:36 <pravins> for indic stuff wildcard is not required presently
05:22:39 <juhp> anyway seems bigger problem is ibus-table performance
05:22:44 <juhp> oh
05:22:46 <juhp> ok
05:23:08 <juhp> pravins: so what does it need?
05:23:32 <pravins> juhp: different algorithm
05:23:46 <pravins> 1) mapping file for indic inscript
05:23:58 <juhp> pravins: scim-tables is using a different algorithm to ibus-table?
05:24:07 <pravins> juhp: no
05:24:07 <juhp> aha
05:24:29 <pravins> only common thing with indic and ibus/scim table is searching in database
05:24:30 <juhp> ok you want to add support for indic maps :)
05:24:48 <pravins> but there are something comes in-between before querying to database
05:25:01 <pravins> those are customization
05:25:11 <juhp> ok I understand now
05:25:19 <juhp> thought maybe that was in the data
05:25:39 <juhp> pravins: so why is scim-tables better than ibus-table?  just performance?
05:25:52 <pravins> juhp: yes, performance
05:25:56 <juhp> ah
05:26:00 <pravins> size of databse
05:26:16 <pravins> ease of adding add removing phrases from user side
05:26:17 <juhp> pravins: cos originally ibus-table was not working at all?? i thought
05:27:01 <juhp> seems I am quite confused...
05:27:09 <pravins> juhp: even me too :)
05:27:12 <juhp> heh
05:27:15 <tagoh> pravins: if "querying to database" is common, doesn't that mean database is also common no? wonder why the size of the database is a matter here then.
05:27:20 <pravins> hope so with alpha we will be in good shape
05:27:37 <juhp> tagoh: the binary format are completely different I think
05:27:39 <juhp> formats
05:27:53 <pravins> yeah
05:27:58 <juhp> pravins: so can you please clarify - ibus-table works or not now?
05:28:02 <tagoh> juhp: aha. can't it use the same format?
05:28:22 <juhp> tagoh: ibus-table uses a real db (sqlite?)
05:28:34 <juhp> tagoh: perhap
05:28:34 <juhp> s
05:28:58 <tagoh> heh. but it's not a good performance right? :)
05:29:06 <juhp> seems not...
05:29:15 <pravins> juhp: it is working, but need some customization for speed and efficiency
05:29:24 <juhp> tagoh: right so db might be bottleneck
05:29:28 <juhp> okay
05:29:32 <juhp> finally got it
05:29:36 <pravins> juhp: yeah ibus-table is swlite
05:29:44 <pravins> s/swlite/sqlite
05:29:51 <tagoh> so the question is why not using the same format in ibus-table
05:29:57 <juhp> tagoh: right
05:30:04 <juhp> we should ask upstream
05:30:09 <pravins> tagoh: yeah, we really required that from long-term
05:30:43 <tagoh> sounds like needing more discussion for that
05:30:55 <juhp> it should not be that hard to port the scim-tables binary format to ibus-table I suppose
05:30:58 <juhp> yeah
05:32:01 <juhp> any bugs to look at for F15?
05:33:16 <tagoh> getting more bugs last week: guess most of them are related to RupeeSign or package reviews.
05:33:54 <juhp> pravins: so we could ship booster as in, in principle?  how bad is the ibus-table performance?  unusable?
05:33:59 <juhp> as is
05:34:46 <pravins> juhp: its taking some second after first key press
05:34:56 <juhp> pravins: and then?
05:35:06 <juhp> only once?
05:35:06 <pravins> after second press it is ok
05:35:17 <juhp> and for next input?
05:35:27 <pravins> like, if we press "k" from k there are so many words starting
05:35:37 <pravins> so it load all them in candidate
05:35:40 <pravins> so it is slow
05:35:52 <tagoh> does it happen every time?
05:35:53 <juhp> pravins: so only very first time it is slow?  next word is ok?
05:35:56 <pravins> after "ka" phrases become little
05:36:00 <pravins> less
05:36:18 <juhp> next input word?
05:36:55 <juhp> pravins: ?
05:36:59 <pravins> for each word, initially it takes time
05:37:02 <juhp> oh
05:37:06 <juhp> I see
05:37:16 <pravins> juhp: tagoh i can send you algorithm wait
05:37:51 <juhp> anyway so current ibus-table is unusable then for booster
05:38:33 <juhp> tagoh: sorry probably we should move on
05:38:41 <tagoh> so what's the plan to go so far?
05:38:54 <pravins> sent mail
05:39:02 <juhp> guess we need to improve the ibus side
05:39:39 <juhp> pravins: but anyway just forking is not going to improve the performance - need to change format perhaps
05:39:55 <pravins> tagoh: taking ibus-table adding some initial conditions before querying db
05:40:08 <pravins> juhp: i agree
05:40:12 <tagoh> sure. that would be nice if we could use the same format to scim-table: if it gives it a good performance.
05:40:13 <juhp> but I think scim-tables may load whole table to memory whereas ibus-table just read the db probably
05:40:49 <tagoh> ok, let's move on
05:40:57 <tagoh> #topic Package Updates
05:40:59 <tagoh> anything?
05:42:13 <tagoh> shall we move on?
05:42:23 <tagoh> #topic Fonts and Rendering
05:42:56 <tagoh> no particular response from behdad for new fontconfig. fyi
05:43:41 <tagoh> as I said at the beginning of this meeting, some fixes may be postponed to f16 perhaps since it requires new features in fontconfig
05:44:22 <tagoh> how about any progress on harfbuzz-ng or fonts?
05:44:34 <juhp> aha
05:46:38 <tagoh> ok, better move on.
05:46:43 <tagoh> #topic Input Method
05:47:29 <tagoh> not applying imsettings on gomme3 has been fixed in imsettings-1.1.0-1.fc15.
05:48:36 <tagoh> I presumed that it still works since gconf has dconf backend. but seems not. so added new backend for gnome3.
05:50:04 <tagoh> anyway, any updates for IM?
05:50:12 <juhp> just saw again the liberation-fonts bug about croscorefonts
05:50:30 <tagoh> bug#?
05:50:39 <juhp> sorry missed the topic
05:50:50 <juhp> https://bugzilla.redhat.com/show_bug.cgi?id=637606
05:51:24 <juhp> just added a provocative comment... wondering they might make a replacement
05:51:39 <juhp> but I don't know enough about the coverage
05:51:58 <juhp> ?
05:52:08 <juhp> pravins: any thoughts?
05:53:06 <tagoh> aha
05:53:10 <juhp> anyway generally feel we don't have enough resources to really maintain Liberation fonts as well as we would like
05:53:38 <pravins> juhp: right, those are many characters
05:54:07 <pravins> if only few characters then we can add that easily, having a professional font designer will help lot
05:54:20 <juhp> yeah looks like they have a lot more glyphs compared to what has been added to Liberation
05:54:47 <juhp> but probably a few eyebrows would be raised
05:55:08 <juhp> would be good to compare the coverage anyway
05:55:36 <pravins> might be we can reported
05:55:39 <tagoh> what's the issue to improve Liberation? no resources to work on?
05:55:48 <juhp> fujiwarat: I filed a bug earlier about dropping the Alt_Gr+Release hotkey from ibus
05:56:13 <juhp> tagoh: mostly yeah and design expertise
05:56:19 <fujiwarat> OK, I see.
05:56:45 <tagoh> juhp: aha
05:57:17 <fujiwarat> Is it critical for f15?
05:57:46 <juhp> fujiwarat: it would be good IMHO - I don't think it is a blocker - though might be for Indic ;o)
05:58:25 <juhp> please tell upstream that using a modifier for a hotkey is a Bad Idea ;o)
05:58:55 <fujiwarat> OK, I see. thx.
05:59:39 <tagoh> ok, anything else?
05:59:55 <juhp> fujiwarat: I can file an upstream bug if you think it helps
06:00:07 <juhp> but mostly want it fixed in fedora anyway
06:01:32 <fujiwarat> ok
06:02:54 <tagoh> no more topics to discuss in the meeting? otherwise let's close the meeting shortly.
06:05:28 <tagoh> ok, thanks everyone for the meeting! if missing anything, you could add that to the next agenda or bring it up on #fedora-i18n, keeping discussions there.
06:05:33 <tagoh> #endmeeting