[01:09] <chrisccoulson> mdeslaur - we maintain the gnome-screensaver packaging in bzr btw (just for future reference)
[01:09] <chrisccoulson> i was just about to upload a new version and noticed the bzr branch was out of date ;)
[07:49] <and471> mvo:hi
[07:49] <mvo> hey and471!
[07:50] <mvo> and471: I merged a history file writing into apt - I'm sure you will like that :)
[07:50] <and471> mvo: yay!
[07:50] <mvo> and471: I will upload it today (doing some final tests, but should be fine)
[07:51] <and471> mvo: what is the basic format of the history?
[07:52] <mvo> https://wiki.ubuntu.com/FoundationsTeam/Specs/LucidSoftwareCenterHistory
[07:52] <mvo> and471: a simple rfc822 format, the python-debian package with the debian_bundle.deb822 parser should cope with it just fine
[07:53] <mvo> the only complication is that it will do log rotation and keep .gz copies of the chnages
[07:53] <mvo> and471: it would be cool to code the log viewer in a way that it could be used standalone as well (if that is possible with not too much effort)
[07:54] <and471> mvo: hmm interesting... :-)
[07:54] <mvo> :)
[07:54]  * mvo is pretty excited about the feature
[07:56] <and471> mvo: once it is uploaded tell me and I shall begin hacking :-) though I may need a demo of how the parsing with debian_bundle.deb822 works
[08:03] <and471> mvo: can I ask you a question?
[08:03] <mvo> sure
[08:04] <and471> mvo: I have a goocanvas.Canvas and some items on that canvas, how would I set up dragging and dropping of those items?
[08:04] <and471> mvo, (within the canvas) so not gtk. drag and drop stuff, but button release events etc.
[08:06] <mvo> and471: I have no experience with that, but I'm happy to have a look (it sounds fun!) - is the code in a branch somewhere?
[08:06] <and471> mvo: sure I shall just push it
[08:06] <mvo> and471: it means you have to implement d-n-d "by-hand" because its not there is GooCanvas yet?
[08:07] <and471> mvo: no I am trying to drag and drop items within goocanvas (ie. not drag them to another widget)
[08:08] <and471> mvo, it is for a flowchart creator I have starting creating :-) https://code.launchpad.net/~rugby471/shauri/trunk
[08:08] <mvo> oh, that sounds like you want to draw it every time on mousemove
[08:08] <and471> mvo: the canvas stuff is in shauri/view/canvas.py (you will notice the structure of the project looks similiar :-] )
[08:08] <mvo> and delete (and restore the bg) of the previous position
[08:08] <and471> mvo, yup
[08:08] <mvo> heh :)
[08:08] <mvo> nice
[08:10] <mvo> I guess its "just" that, keeping track on on-mouse, the old cursor position, the new and lots of drawing (and deleting).
[08:10] <mvo> I'm sure there are nasty corner cases and such too :)
[08:11] <and471> mvo: what is the gobject signal(s) I need to track?
[08:11] <and471> is > are
[08:13] <mvo> and471: I think you need motion-notify-event on the canvas itself. if there is a mouse-click on a item, go into drag-state, if the mouse is moved (motifion-notify-event) in that state, do the drawing magic. when the mouse is released, end that state
[08:14] <mvo> you should get a event that gives you "event.x", "event.y"
[08:14] <and471> mvo: ok cool thanks :-)
[08:14] <mvo> (GdkEventMotion)
[08:14] <mvo> good luck! and let me know how that goes
[08:15] <and471> mvo: thanks I shall :-) I was trying to make a flowchart the other day, I installed and opened up dia and thought "this is too hard" ;)
[08:15] <and471> mvo, this is going to be a really simple and easy-to-use program :-)
[08:16] <pitti> Good morning
[08:16] <mvo> and471: cool!
[08:16] <mvo> hey pitti
[08:16] <and471> pitti, hello
[08:31] <chrisccoulson> good morning everyone
[08:32] <pitti> hey chrisccoulson
[08:33] <chrisccoulson> hey pitti, how are you today?
[08:33] <chrisccoulson> are you winding down for the festive break yet? ;)
[08:34] <pitti> I'm great, how are you?
[08:34] <pitti> I'm still at work until tomorrow
[08:35] <and471> mvo: okay I have got it working however I have one problem, the gd.Event gives me the x and y coordinates of the window not the widget, is there anyway to convert these?
[08:35] <and471> *gkd.Event
[08:35] <chrisccoulson> yeah, i'm ok too. i got most of my christmas grocery shopping done last night, which was not fun, but i'm glad it's out of the way now :)
[08:36] <and471> *gdk.Event
[08:39] <and471> mvo, don't worry I ahve go it working, the gdk.Event does give me the correct coordinates
[08:43] <mvo> and471: aha, cool
[08:45] <and471> mvo: yay it is working :-)
[08:45] <mvo> sweet!
[09:00] <chrisccoulson> pitti - do you just want me to just upload the fix for bug 428884 to karmic-proposed, or do you want to have a quick look at the fix beforehand? FYI, the patch i uploaded to lucid last night is http://bazaar.launchpad.net/~ubuntu-desktop/gnome-screensaver/ubuntu/revision/48#debian/patches/10_legacy_scrsvr_inhibit.patch
[09:00] <chrisccoulson> unfortunately, it requires an extra build-depend on libxtst-dev to work, but that's already required by other desktop components anyway
[09:02] <pitti> chrisccoulson: new b-dep in main doesn't sound like a problem
[09:02] <chrisccoulson> pitti - thanks
[09:02] <pitti> chrisccoulson: would there be a way to avoid patching autoconf stuff for the SRU?
[09:02] <pitti> and just assume that we have it?
[09:03] <pitti> the Makefile.am change can probably be represented as a single-line diff in Makefile.in
[09:03] <pitti> I don't insist on it, but it might make the change much smaller and less error prone
[09:04] <chrisccoulson> pitti - yeah, i can probably do that. we'd still need to keep the Makefile change to correctly link against libxtst though, but i can remove the configure check for the SRU. i only added that to make it more acceptable for upstream
[10:04] <chrisccoulson> vuntz - we had a user report a gnome-session bug because they tried to get nautilus to open a window on login, by adding a nautilus autostart file in gnome-session-properties, but it didn't work because they ended up with 2 applications with the same app_id (and only one of them is started). would you consider that to be a bug, or something which is unlikely to change? (i'm wondering whether to close the launchpad bug or not)
[13:06] <huats> hello everyone !
[13:08] <mpt> mvo, hi, I thought you were supposed to be on holiday ;-)
[13:10] <mvo> mpt: yesterday, today and tomorrow I'm at work :)
[13:10] <mpt> mvo, ah, I didn't call you yesterday because I didn't see you online anywhere
[13:13] <mvo> mpt: right, I was on vac yesterday
[13:14] <mvo> but not today and tomorrow
[13:16] <mpt> mvo, I just updated trunk, and it's in an interesting state -- all the application names and icons are gone
[13:19] <mvo> mpt: oh? does "rm data/xapian/*" help?
[13:20] <mpt> oh, I wasn't paying attention to what was happening in the terminal
[13:21] <mpt> it was full of raise IndexError("No app '%s' for '%s' in database" % (appname,pkgname)) etc
[13:21] <mpt> yes, all fixed now
[13:22] <mvo> cool, thanks
[13:28] <chrisccoulson> good afternoon everyone
[13:37] <mpt> mvo, I'm thinking about how to show non-application packages in search results
[13:37] <mpt> mvo, and one example I'm using is "the" (The Hessling Editor)
[13:39] <mvo> mpt: we can/could force exact package name searches as part of the query
[13:39] <mvo> that would match "the"
[13:39] <mpt> yes
[13:40] <mpt> So, something like: If there's a package that exactly matches, show that first, then show all application matches alphabetically, and hide non-application matches behind a toggle of some sort
[13:40] <mpt> (other non-application matches, I mean)
[13:41] <mpt> ... except in "Tech Stuff", where non-application matches should be shown by default
[13:42] <mpt> Currently searching in USC for "hessling" returns the, but searching for "the" does not
[13:42] <mvo> mpt: I'm not sure about the alphabetic, I send you a environment that you can use to modify the behavior a while ago
[13:43] <mvo> mpt: and alphabetic is confusing if you search for something with "z" in it IMO
[13:43] <mpt> maybe
[13:43] <mvo> mpt: I thnk I send it to you by mail, just play with it to get a feeling for the sorting
[13:43] <mpt> Another problem with alphabetic is where we get wildly popular and ISVs start gaming it by starting their application names with underscores or "AAA"
[13:44] <mpt> (that happened with Apple's store)
[13:45] <mvo> oh, so that is the way the apple store is doing it by default? interessting
[13:45] <mpt> I think they stopped, I'm not sure
[13:45] <mpt> haven't used it myself :-)
[13:45] <mpt> http://arstechnica.com/apple/news/2008/07/app-store-alphabetical-listings-quietly-fixed.ars
[13:46]  * mvo nods
[13:47] <mvo> mpt: we currently implement three different sortings, if you can't find the environemnt vars I can look them up for you again (sorry, I don't have it right now)
[13:55] <mpt> mvo, sorry, I haven't found that message (I went through all messages where you mentioned "sort")
[13:57] <mvo> mpt: try "PYTHONPATH=. SOFTWARE_CENTER_SEARCHES_SORT_MODE=popcon ./software-center"
[13:57] <mvo> mpt: it supports "xapian"
[13:57] <mvo> mpt: alphabetic
[13:57] <mvo> and popcon
[13:58] <mpt> hm, searching is hard
[14:00] <mpt> None of those options, when I search for "python", return any version of the Python interpreter as #1 or #2
[14:00] <mpt> I guess we can get around that partly with exact matches
[14:01] <mpt> but that might not help with searches like "openoffice"
[14:05] <mvo> yeah, search is hard, espcially when it neeeds be fast and exact. give me a minute to look if I can get the package name matches mixed into the search
[14:40] <mvo> mpt: please update and check how well the xapian mode now works. I added scaling for pkgname and a strong favor towards applications
[14:41] <mvo> r477
[14:43] <pitti> hm, any gnome committer here who could apply an approved patch for me?
[14:43] <mvo> but popcon (by its nature) gives us very good results for common searches
[14:43] <mvo> mpt: its hard to get the exact behavior you want, there is no simple "do-that search first and append the other results later"
[14:44] <pitti> mclasen: thanks for the glib assert patch review; I can't commit myself, but I'll find someone
[14:44] <mclasen> pitti: we'll get it in, no worries
[14:44] <mpt> mvo, one lookup and one search, splice the results together? :-)
[14:45] <mclasen> pitti: now I need to go and find some abrt hacker; I think they're all off for xmas already...
[14:45] <pitti> mclasen: that problem sounds familiar :)
[14:46] <pitti> OTOH these three days are great for getting stuff done and reduce the backlog
[14:46]  * pitti has $work inbox down to three
[14:47] <mvo> mpt: right, thats like saying "just return the right results"
[14:47] <mpt> mvo, why is that? Is it because it would take twice as long as doing just one search?
[14:51] <mvo> mpt: its because the bit that displays the application gets a query as input. it does not care much if its in search, display (sub)category or search-in-category(or sub-category) mode. its a nice abstraction. that bits needs to be redone to take multiple queries and stitch them together. maybe hard is the wrong word, but it does not map very well currently
[14:52] <mpt> mvo, is that something you'd need to rewrite anyway to hide non-application packages from most search results by default?
[14:52] <mvo> mpt: no, that is relatively easy to do by adding a "AND" or "AND_NOT" applications
[14:52] <mpt> ok
[14:55] <mpt> hm, something must be broken
[14:55] <mvo> mpt: with the recent commit?
[14:57] <mpt> mvo, in r477 I search for "thunderbird" and I get 200 results: Mozilla Thunderbird is #56, and most of the others don't mention "thunderbird" at all
[14:59] <mpt> mvo, http://paste.ubuntu.com/344756/
[15:00] <mvo> mpt: with SEACHES_SORT_MODE=xapian ? or with the default mode?
[15:01] <mpt> mvo, default mode
[15:01] <mvo> SOFTWARE_CENTER_ ... =xapian
[15:01] <mpt> Was that change you made just to the xapian mode?
[15:01] <mvo> mpt: that may well be broken now :/ please check the xapian one
[15:01] <mpt> ok
[15:05] <mpt> mvo, http://paste.ubuntu.com/344759/
[15:05] <mpt> It's improved on non-application packages, worsened on applications
[15:08] <mpt> It's also consistently returning ~200 results where most of them don't mention the search string at all
[15:10] <mvo> hm, hm, so a different approach is needed
[15:11] <mpt> actually, I need to increase the sample
[15:11] <mpt> it's doing fairly well on other things I'm trying
[15:11] <mpt> but getting rid of those completely irrelevant results would be nice anyway :-)
[15:15]  * mvo nods
[15:19] <mpt> mvo, http://paste.ubuntu.com/344765/
[15:22] <mpt> Emilia Pinball seems to come up in the top 20 for just about every search I do :-)
[15:22] <mpt> e.g. it's the #12 result for "spreadsheet"
[15:28] <mpt> mvo, so on balance, from that sample I think r477-xapian is a bit worse than r476
[15:46] <cassidy> kenvandine, hey! Any chance to see https://bugs.edge.launchpad.net/ubuntu/+source/telepathy-gabble/+bug/494591 fixed in Karmic?
[15:46] <mvo> mpt: hm, I just looked into a custom weight implementation to do custom relevance raking with xapian, but that seems to be not possible with the python bindings. I think it would solve the problem neatly. oh well
[15:46] <mvo> s/would/would have/
[16:18] <mvo> mpt: please try r478, still not perfect, but I think better
[16:34] <mpt> mvo, with explicit xapian mode, or default mode?
[16:36] <mvo> mpt: default, sorry for not mentioning that
[16:42] <mpt> mvo, http://paste.ubuntu.com/344818/
[16:42] <mpt> r479 does a bit better than r476
[16:43] <mpt> r479 in xapian mode does better still.
[16:43] <mvo> mpt: that is interessting, dive into python is a problem to become #1 because python is also a package name and huge popularity
[16:45] <mpt> And in both default and xapian mode, I don't see any completely irrelevant results, nice work
[16:45] <mvo> thanks
[16:46] <mpt> mvo, shouldn't it matter that the python package doesn't mention "dive" at all? :-)
[16:49] <mvo> mpt: the package names are currently a OR search, I can make that a AND, but that will be on the expenses of other queries. but if you prefer that that is fine with me (OTOH there is no dive AND in AND python package)
[16:50] <mpt> mvo, sorry, I don't understand what you mean by "there is no dive AND in AND python package"
[16:51] <mvo> mpt: I mean, for the package name query, there is no exact match there. but yeah, for the other one I need to check what is going on, but xapian is sometimes still magic
[16:52] <mpt> mvo, do you mean that with an AND search, "dive into python" would return no results?
[16:52] <mvo> mpt: for a AND package name search yes, it would be a "dive AND into AND python" on the pkgname
[16:53] <mpt> Well, the package name isn't "dive" or "into" or "python" anyway
[16:53] <mpt> So in both cases it's looking at the description instead, right?
[16:54] <mvo> yes
[16:54] <mvo> hm, but its a good question why its not doing a AND there
[16:54] <mpt> We're hampered here by the synopsis being "free Python book for experienced programmers", rather than "Dive Into Python, a reference for experienced programmers" or something like that
[16:55]  * mvo nods
[16:55] <mpt> Unlike Google and Bing developers, we have the advantage of being able to improve our dataset :-)
[16:55] <mvo> indeed
[16:56] <mvo> user-contributed-metadata ftw
[16:56] <mpt> exactly
[16:56] <mpt> In the meantime, the latest xapian mode looks like a good default to be going on with
[16:56] <mvo> ok, cool
[16:57] <mpt> though I'm happy to test something where the package search is ANDed
[16:57] <mpt> package name search, I mean
[16:59] <mvo> mpt: ok, cool. I need to leave now (appointment this evening) - but I can commit a patch for this tomorrow morning - just ping me when you are online and I'm happy to do that
[16:59] <mpt> mvo, ok, or put it in a branch and mail me, whichever suits you
[16:59] <mpt> thank you for your attention to detail
[17:00]  * mpt wonders what package should come up #1 in a search for "realtime kernel"
[17:00] <mpt> linux-image-rt, I guess
[17:01]  * mvo waves
[17:01] <mpt> tchau
[17:02] <mpt> But there we have crap data because the package info doesn't mention "realtime" at all
[17:02] <mpt> "This package will always depend on the latest rt kernel image available." </nothelpful>
[17:13] <crimsun> mpt: I'll patch that, thanks
[17:14] <mpt> crimsun, I'm part-way through writing it up, would you like me to hit the "Submit Bug Report" button, or would that be more bothersome than useful?
[17:14] <crimsun> submit, and please let me know the bug report #
[17:15] <crimsun> thanks
[17:17] <mpt> crimsun, reported bug 499524, thank you kindly
[17:18] <mpt> crimsun, btw, do you know why <https://launchpad.net/ubuntu/+search?text=linux-image-rt> shows the linux-image-rt binary package coming from two different source packages for the same series? Are they for different architectures, or what?
[17:23] <crimsun> mpt: originally targeted to two different base kernels
[17:23] <crimsun> 2.6.31 and 2.6.32 bases
[17:25] <mpt> hm
[17:25] <mpt> When I search Synaptic for "linux-image-rt" I get one result, but when I search Ubuntu Software Center I get two identical results
[17:45] <crimsun> mpt: would it be acceptable to use "Realtime" instead of "real-time"?
[17:45] <crimsun> mpt: upstream uses the former
[17:46] <mpt> crimsun, heh, I was just wondering how to finagle the description so that it would contain both :-)
[17:46] <mpt> so that a search for either would return it
[17:47] <crimsun> well, technically it's "Realtime Linux" and "RT" both, but I agree that we need to avoid confusing users with other "rt"-like packages
[17:48] <bjsnider> is it acceptable for a package to wipe out a userland config file as part of an update, or is it considered to be destruction of data?
[17:48] <crimsun> i.e., we don't want people to mistakenly think they're getting "Real Time Linux" (i.e., RTAI)
[17:49] <mpt> crimsun, so are some/most of the mentions of "real-time kernel" on the Web people referring to RTAI?
[17:49] <crimsun> I'll just make this clearer in the longer description
[17:49] <mpt> http://googlefight.com/index.php?lang=en_GB&word1="real-time+kernel"&word2="realtime+kernel"
[17:49] <mpt> ok
[17:51] <crimsun> "real-time kernel" returns mostly "Realtime Linux" (i.e., Ubuntu's linux-meta-rt-related) results
[17:51] <crimsun> bjsnider: note the distinction between conffile and configuration file
[18:00] <bjsnider> crimsun, let me be specific. ~/.gnome2/backgrounds.xml
[18:01] <bjsnider> that file is created if the user ever changes anything about the default backgrounds such as their colour
[18:01] <bjsnider> it overrides the xml files in /usr/share/backgrounds
[20:37] <dobey> what's the installer project name? ubiquity?
[20:37] <Keybuk> the Live installer, yus
[20:39] <dobey> cool, thanks
[22:06] <chrisccoulson> pitti - i've uploaded the gnome-screensaver SRU now