[10:42] <lool> asac: Hey
[10:42] <lool> asac: How are you doing?
[10:42] <lool> asac: So, let's move the mbf discussion here
[10:42] <lool> asac: I think the SVG issue might be specific to 1.8
[10:43] <lool> So perhaps we should try moving to 1.9 before investigating
[10:43] <lool> asac: So, I think you said we could either write a xul app or fix the security checks
[10:43] <lool> asac: Did you hear from upstream about them?
[10:44] <lool> asac: I think you traced the problem of xul 1.9's mbf to be some security nest clearing the javascript vars before being passed to the html
[10:44] <asac> lool: we could do a xul application out of this mobile-basic-flash thing.
[10:45] <asac> not yet sure how to best interface with the hildondesktop API though
[10:45] <lool> asac: Ok
[10:45] <asac> lool: how does mbf interact with hildon desktop?
[10:45] <asac> does it just send "start X", start Y?
[10:46] <lool> asac: So instead of setting up a gtk_mozembed thing, it would create a xul thing?
[10:46] <asac> lool: basically yes.
[10:46] <asac> lool: i have to look at the code
[10:46] <asac> let me get that
[10:47] <lool> asac: It basically receives events from being a HD plugin I think, and it triggers launching of app
[10:47] <lool> Like "app foo starting"
[10:47] <lool> (from a quick grep)
[10:47] <lool> asac: http://ppa.launchpad.net/ubuntu-mobile/ubuntu/pool/main/m/mobile-basic-flash/mobile-basic-flash_0.44-0ubuntu4.dsc
[10:48] <lool> asac: Do you think we should start from the moblin (tip) version?
[10:48] <asac> lool: what kind of events does it need to deal with?
[10:49] <lool> asac: From hildon?  not many I think
[10:49] <asac> yes from hildon
[10:49] <lool> Things like X events or home plugins change notifications
[10:49] <lool> asac: But we could simply keep all of this in the current C code, no?
[10:50] <lool> asac: Hmm perhaps we should think carefully at which version we port
[10:50] <zamba> is it possible to install ubuntu-mobile on a sony ericsson p1i?
[10:50] <lool> asac: Currently there are three main trees, 1) moblin 2) UME 3) MSG
[10:50] <lool> asac: MSG people had to bridge sound events triggerred by actions in their flash movie
[10:51] <lool> asac: Do you have access to their tree?  they changed it a whole lot
[10:52] <lool> asac: So perhaps the safest is to take tip, and then port to downstreams?
[10:53] <lool> And moblin's tree is at moblin.org:/home/repos/projects/mobile-basic-flash.git
[10:53] <asac> lool: i have to think about it
[10:53] <asac> cant tell whats really the best way to go on
[10:54] <lool> asac: Ok; what about the SVG issue?  Did SVG work fine in 1.8?
[10:54] <asac> lool: afaik it works in firefox 2
[10:54] <lool> asac: Basically MBF calls gtk to get the icon pathname
[10:54] <asac> so it should work in 1.8
[10:54] <lool> Then inserts the HTML snippet in the DOM with an <img> pointing at it (I guess)
[10:55] <lool> This works for pngs, but with svg it displays once, disappears, displays a second time, disappears and I'm in an infinite loop somewhere
[10:55] <asac> lool: interesting
[10:55] <asac> lool: opening that pic in firefox-2 works fine?
[10:56] <lool> asac: Dunno
[10:56] <asac> thats certainly a thing you could try
[10:56] <lool> (Installing it)
[10:57] <asac> lool: backup your production firefox profile to prevent eventual mess when switching back and forward
[10:57] <asac> (2.0 <-> 3.0)
[10:58] <lool> I'm installing it on my Q1 where I don't run firefox
[11:00] <lool> asac: Works fine from firefox-2
[11:01] <lool> But then the HTML might cause the SVG to be resized or something
[11:02] <lool> the image is loaded with a background-image: css property
[11:03] <asac> hmm
[11:03] <asac> might make a difference
[11:21] <lool> asac: I tried reproducing with      <object data='/usr/share/icons/hicolor/scalable/apps/update-manager.svg' width='48' height='48' ></object>
[11:22] <lool> To no luck yet
[11:22] <asac> lool: didnt you say that mbf is using <img tags?
[11:23] <lool> It was just a guess
[11:23] <lool> Looking at the actual html, it uses <object data="">
[11:34] <lool> asac: All my test from firefox-2 work fine :-(
[11:35] <lool> I even have the javascript working and all
[11:44] <lool> asac: Oh, it doesn't work at all in my firefox 3 here
[11:45] <asac> http://www.w3schools.com/svg/svg_examples.asp
[11:45] <asac> those work for me
[11:45] <lool> asac: SVG works
[11:45] <lool> asac: I mean the javascript of mbf
[11:45] <asac> yeah
[11:45] <lool> It works fine with SVG loading and all in firefox-2
[11:46] <asac> thats why its broken with xulrunner 1.9 :)
[11:46] <lool> (it hangs mbf)
[11:46] <lool> asac: So in 1.9, it seems desktop_mgr.js is the culprit
[11:46] <lool> asac: Did you pinpoint where?
[11:46] <asac> lool: afaict its everly load_url("javascript:...") that is not working
[11:47] <asac> in the .cpp code
[11:47] <asac> what does desktop_mgr.js do?
[11:47] <lool> The <script type="text/javascript" src="grid_home.js"> are rejected too
[11:47] <lool> In fact it's not specific to the .js
[11:48] <lool> Missing </script>!
[11:49] <lool> This works   <script type='text/javascript' src="desktop_mgr.js"/></script>
[11:49] <asac> lool: cool does it fix it?
[11:49] <lool> I have no idea, I don't have the xul 1.9 build
[11:49] <lool> If you have a 1.9 xul mbf binary, I could try it out
[11:50] <lool> asac: http://people.ubuntu.com/~lool/mobile-basic-flash/home.html
[11:51] <asac> let me check master
[11:51] <asac> i think the xul 1.9 changes are already committed there
[11:51] <lool> They are
[11:52]  * lool grabs xul1.9-dev
[11:53] <asac> lool: you need to patch configure.ac
[11:53] <asac> http://paste.ubuntu.com/19856/
[11:53] <lool> asac: Ok; git log says they applied your patch
[11:53] <lool> Right, that looks more like it
[11:54] <asac> lool: oh ... no all is fine i think
[11:54] <lool> Ah, I applied your patch
[11:55] <lool> Ok: reverted; autogen
[11:55] <asac> lool: good
[11:55] <lool> git is so cool
[11:55] <asac> hush
[11:55] <lool> git clean -df; git reset --hard HEAD :)
[11:56] <asac> i just used git checkout configure.ac :)
[11:56] <asac> but resetting is nice ... especially that you can commit -c ORIG_HEAD :)
[11:56] <lool> asac: It's empty, unfortunately
[12:00] <lool> asac: Anyway, I think bspencer demonstrated how lower level calls were failing already
[12:00] <lool> What I spotted was just another bug, later down, but it's not the xul 1.9 blocker :-/
[12:00] <asac> lool: right. its the load_url("javascript..." thing that doesnt work
[12:01] <lool> Yeah
[12:03] <asac> lool: i would instantly debug this if we could make a plain gtk window out of it and put the mbf widget in it
[12:03] <asac> so i can drop the whole hildon stuff and reduce the testcase to a developer friendly complexity
[12:04] <lool> asac: Ok
[12:08] <asac> lool: maybe we can just make "home_plugin" a GtkWindow?
[12:08] <lool> That's what I'm looking at
[12:08] <lool> MobileBasicHomePlugin could be instanciated
[12:09] <asac> (in _plugin_init i mean)
[12:09] <lool> Good idea, perhaps you can try that
[12:09] <asac> hmmm the MobileBasicHomePlugin is a custom widget from what i can tell
[12:09] <lool> I'm looking at home_screen_changed(), it seems to do the init we want
[12:10] <asac> ah no ... i am wrong ... i looked at wrong struct. its derived from HildonDesktopHomeItem
[12:11] <lool> I think we could change home_screen_changed() to use a GtkWindow instead of a GdkWindow
[12:12] <asac> lool: the mozembed widget is added to home_plugin in plugin_init
[12:13] <asac> so i think just changing home_screen_changed is not enough
[12:22] <asac> lool: http://paste.ubuntu.com/19864/
[12:22] <asac> thats ugly but it starts a window somehow
[12:26] <asac> ok you would also require a hacky patch like http://paste.ubuntu.com/19865/ to test in-source html
[12:26] <lool> asac: I'm doing the same thing ATM :)
[12:26] <asac> they should honour ATM :)
[12:26] <asac> ouch
[12:26] <asac> honour DATADIR
[12:26] <asac> or something
[12:26] <lool> http://paste.ubuntu.com/19866/ is what I was doing
[12:27] <lool> funny how we have basically the same thing
[12:27] <asac> lool: did you switch to GtkWindow parent class too?
[12:27] <lool> No, I was doing that
[12:28] <asac> for me its all black now
[12:29] <asac> is that the bug?
[12:29] <lool> asac: i'll have to drop off for lunch now; since you're further than I have with a GtkWindow displayed, could you continue investigating? :)
[12:29] <lool> asac: Yes that's the bug
[12:29] <lool> asac: You get any stderr output?
[12:29] <asac> lool: http://paste.ubuntu.com/19867/
[12:29] <asac> export NSPR_LOG_MODULES=all:5
[12:29] <asac> will show that
[12:30] <asac> so its obvious. gtkmozembed doesnt support javascript: in load_url anymore
[12:30] <lool> asac: Good stuff
[12:30] <lool> asac: Are you sure it's not the function missing?
[12:31] <lool> asac: Did you change the <script src=""/> into <script></script>?
[12:31] <asac> let me check if there is some error before
[12:31] <lool> Cause that was blocking my firefox (xul 1.9) to render the page I handed you
[12:32] <asac> lool: it works ;)
[12:32] <asac> it was the script thing
[12:32] <lool> asac: Cool
[12:32] <asac> lool: http://paste.ubuntu.com/19868/
[12:32] <lool> Yeah
[12:33] <asac> but with that patch its not working for you?
[12:33] <asac> i mean the real build?
[12:33] <lool> asac: I was a bit hackish: I built from mbolin git and copied the .so over my installed plygin
[12:33] <lool> Perhaps I misbuilt it
[12:33] <lool> asac: Could you rebase the patch atop the .dsc I handed you?
[12:34] <lool> http://ppa.launchpad.net/ubuntu-mobile/ubuntu/pool/main/m/mobile-basic-flash/mobile-basic-flash_0.44-0ubuntu4.dsc
[12:34] <asac> lool: no ... that thing is not xul 1.9 ready
[12:35] <asac> we need  master from what i can tell
[12:35] <asac> and the xul 1.9 patch cannot be easily rebased. i would have to redo it i think
[12:35] <asac> better git merge master on the packaging tree i guess
[12:36] <asac> let me see how far i get when trying to merge that thing on the hardy branch
[12:36] <lool> Ok; we don't have a packaging git tree sadly, but I'll try to look into that
[12:37] <lool> I really need to cook lunch for my wife now, she's complaining that she's hungry
[12:40] <asac> lool: there is a hardy branch
[12:40] <asac> why not put all the packaging files in there
[12:40] <emgent> heya
[12:40] <asac> afaict its just a few files missing
[12:40] <asac> for instance no debian/changelog ;)
[12:40] <asac> but debian/ dir exists
[12:40] <asac> and debian/rules
[12:42] <asac> ok hardy branch building
[12:43] <asac> lets see
[12:43] <asac> lool: hmmm ... you probably cannot use the binaries from amd64?
[12:44] <asac> lool: pleaes checkout hardy branch and apply http://people.ubuntu.com/~asac/mbf.hardy-to-latest.patch
[12:44] <asac> that should be the latest
[12:44] <asac> changelog version might be wrong
[12:58] <asac> lool: let me know if this works
[12:58] <asac> ill take a break now
[12:59] <asac> lool: if that works, I'll just commit it to moblin i guess
[13:11] <lool> asac: It doesn't apply unfortunately
[13:11] <lool> asac: What do you mean with hardy branch?
[13:11] <lool> origin/hardy oh
[13:12] <lool> I don't have access to this
[13:12] <lool> asac: it's not in sync with the ppa .dsc sadly
[13:12] <lool> I guess they created it when they moved packaging out of master
[13:13] <lool> But we can't ensure that uploaded packages and git branches are in sync unless we have commit access, and we didn't get commit access
[13:13] <lool> asac: In the ppa, the source is still named .c for instance
[13:17] <lool> asac: I'm kicking a build of hardy branch + your patch to at least see whether it works
[13:18] <asac> lool: yes try that branch. i dont see that we can quickly fix 0.44
[13:18] <asac> if we know that it works we can do the work if we require a backport for hardy
[13:18] <asac> but first lets check if it works in theory ;)
[13:18] <lool> I came to the same conclusion
[13:19] <asac> lool: i have commit access
[13:19] <asac> you too?
[13:19] <lool> checking for MOZEMBED... configure: error: Package requirements (mozilla-gtkmozembed-embedding >= 1.8.1) were not met:
[13:19] <asac> lool: xulrunner-1.9-dev is required
[13:19] <asac> for me:
[13:19] <asac>  pkg-config --cflags mozilla-gtkmozembed-embedding
[13:19] <asac> -DXPCOM_GLUE -fshort-wchar -I/usr/include/xulrunner-1.9/unstable -I/usr/include/nspr  
[13:20] <lool> asac: I'm pbuilding my lpia binaries
[13:20] <lool> The hardy branch was bdeping on libxul-dev
[13:20] <asac> lool: ok ... yeah changed build-depends from libxul-dev to xulrunner-1.9-dev
[13:20] <asac> lool: forgot to include that in the diff i guess
[13:20] <lool> drwxrwsr-x  8 nobody    dev-mbf       4096 Jul 26  2007 mobile-basic-flash.git
[13:21] <lool> I'm only in group devel, so basically no commit access
[13:21] <lool> You're in dev-brow dev-ma
[13:21] <lool> asac: I don't think you have commit access anymore
[13:22] <asac> oh ok. they have per project permissions
[13:22] <lool> You have mobile-browser.git and moblin-applets.git
[13:22] <asac> yeah. what is the "all" group?
[13:22] <lool> no idea
[13:22] <asac> ;)
[13:22] <asac> ill ask bob if he wants me to commit the mobile-basic-flash fixes ;)
[13:23] <lool> asac: I think it just WORKED !
[13:23] <asac> cool
[13:23] <lool> I'm not sure I believe it
[13:23] <asac> yay \o/
[13:24] <lool> Uh
[13:24] <lool> libxul0d
[13:24] <asac> lool: he?
[13:24] <lool> asac: It's still using 1.8
[13:24] <asac> hmm
[13:24] <asac> lool: you have the link line?
[13:24] <lool> Oh wait
[13:24] <lool> hardcoded in control
[13:24] <lool> Depends: ${shlibs:Depends}, ${misc:Depends}, libxul0d, libwnck22
[13:25] <asac> lool: ldd src/mobilebasichome  | grep libxu
[13:25] <asac> nothing
[13:25] <asac> lool: yeah ... thats just packaging ;)
[13:25]  * asac fixes that in his git clone
[13:25] <lool> ldd /usr/lib/hildon-desktop/libmobilebasichome.so | grep xul => empty
[13:26] <lool> Once I'll have removed the dep, I'll simply wipe libxul0d and see whether it starts :)
[13:27] <asac> ok i have fixed it
[13:27] <asac> here in git
[13:27] <asac> ill ask bob about committing :)
[13:27] <lool> asac: It's fixed in the ppa's packaging too
[13:27] <asac> lool: ok cool
[13:27] <asac> so all is fine?
[13:27] <asac> (once all is committed upstream?)
[13:28] <lool> asac: Sure, then Bob needs to release a tarball and not upload it to ppa
[13:28] <lool> It's going to be very painful to sync the mobile-basic-flash stuff; it never used a patch system and was forked over and over
[13:29] <lool> Only thing installed is xulrunner-1.9 and it comes up!
[13:31] <lool> asac: So this is good to go
[13:31] <asac> lool: what changes do we have for mobile-basic-flash?
[13:31] <lool> asac: So the only change to complete support was that stupid <script> syntax?
[13:31] <asac> yeah ;)
[13:31] <lool> asac: The problem is I don't quite know
[13:31] <lool> asac: I'll have to diff stuff between git and ppa
[13:31] <lool> It's very painful I can tell you
[13:31] <lool> i did it for other modules, it takes hours
[13:32] <lool> If we had been able to use git, or simply a patch system, we wouldn't have this problem
[13:32] <asac> lool: why dont at least work on your own git tree?
[13:32] <asac> i mean you can push it to rookery or somewhere
[13:32] <asac> or use the bzr git plugin ;)
[13:32] <asac> (read-only yes, but for this purpose it would be good enough i guess)
[13:34] <lool> asac: I inherited that situation
[13:34] <lool> asac: Also, everybody needs to be able to change it
[13:34] <lool> And people found bzr too complex already, so I didn't want to force git or git->bzr usage
[13:35] <lool> So you need a place where Intel and Ubuntu people can commit, where it's easy to change things, and then you need to do the import
[13:35] <lool> (of Ubuntu specific changes)
[13:37] <lool> pmcgowan: When you're around... :)
[13:37] <lool> pmcgowan: We have a fix for the xulrunner 1.9 move
[13:38] <lool> asac: I tried SVG with XUL 1.9
[13:38] <lool> asac: It /works/!
[13:38] <lool> asac: But it blinks
[13:39] <lool> asac: You still have your GtkWindow hack?
[13:40] <lool> asac: Could you change the gtk_icon_theme_lookup_icon() calls to pass (GtkIconLookupFlags)0 instead of GTK_ICON_LOOKUP_NO_SVG and change the gtk_icon_theme_prepend_search_path() in an gtk_icon_theme_append_search_path()?
[13:40] <lool> asac: This should give you SVG support
[13:41] <asac> err
[13:41] <asac> sorry, i cant follow
[13:41] <asac> do you want that hack so you can test?
[13:41] <asac> lool: send mail to bob with you cc'ed
[13:41] <lool> asac: Argh
[13:41] <pmcgowan> lool, cool whats the fix?
[13:41] <lool> asac: Just found a regression on the new MBF
[13:41] <asac> lool: regression?
[13:42] <lool> pmcgowan: Change <script/> to <script></script> in the *.html
[13:42] <lool> asac: I can't change categories
[13:42] <lool> asac: You have a Q1?
[13:42] <asac> lool: yeah, but no good keyboard
[13:42] <asac> so doing things is cumbersome
[13:42] <lool> asac: You can ssh to it!
[13:43] <pmcgowan> lool, simple enough, our html may not even have that issue
[13:43] <asac> lool: hehe
[13:43] <asac> yeah
[13:43] <lool> pmcgowan: The rest of the required changes are in the moblin git
[13:43] <lool> pmcgowan: autotools stuff, renaming, etc.
[13:43] <lool> asac: If you prefer, you can build a VM
[13:44] <lool> asac: That's easy to do nowadays
[13:44] <asac> lool: with virtual box?
[13:44] <lool> asac: https://wiki.ubuntu.com/Testing/Cases/UMEinstall
[13:44] <lool> asac: Even if virtualbox if you like
[13:44] <lool> s/if/with
[13:44] <asac> lool: i wont be able to really work on more things today
[13:45] <asac> lool: i have to do something and then prepare for a different battle of real life ;)
[13:45] <lool> pmcgowan: This will probably enhance hildon desktop startup time (it did for me) and will certainly reduce midbrowser startup time
[13:45] <asac> but I'll be here during weekend
[13:45] <lool> asac: Ah, sad to read this
[13:45] <asac> given that you are usually on irc as well, we can certainly continue then
[13:45] <lool> asac: I'll try to grab bspencer
[13:45] <lool> I'm not sure I'll have much time this WE
[13:45] <asac> lool: yeah ... i send the mail, so you can jump in there i guess
[13:46] <asac> maybe wait if he replies :)
[13:46] <lool> I kind of neglected preparations for the coming baby, and have plenty of things to do in the last days
[13:46] <asac> lool: i suggested Co-Maintainer from Ubuntu to commit to moblin branches
[13:46] <asac> lool: oh. ok. better stay off on the weekend then
[13:46] <asac> lool: baby is more important that Ubuntu-Baby ;)
[13:46] <lool> I'll probably be around a little anyway
[13:46] <asac> s/that/than/
[13:47] <asac> I knew that ;) ... how can't you be around the whole weekend?
[13:47] <lool> Plus, the final is built already :-/
[13:47] <lool> Too bad we didn't find out before final
[13:48] <asac> lool: yeah. lack of time. i asked bob to test in firefox
[13:48] <asac> that would have given us the hint
[13:48] <asac> i never received anything back
[13:48] <lool> asac: Well it's how I found about it
[13:49] <lool> I tested in firefox-2, in midbrowser; the console errors helped me catch the error
[13:49] <asac> right ... thats why i am saying that
[13:49] <lool> I then tried to alert("foo");, and it was obvious
[13:49] <asac> bad luck ;) ... the ideas were there; just nobody did it :)
[13:50] <asac> lool: but we decided that its ok to do the migration in a stable update
[13:50] <lool> asac: Ideally, could you throw your hack for GtkWindow as a patch to mbf?  It would allow to -DMAIN_WINDOW or something to get an executable
[13:50] <lool> asac: Did we decide that?  Hmm
[13:50] <lool> Probably we did
[13:50] <asac> lool: yes. two weeks ago or so in mobile meeting
[13:50] <asac> the question came up: "is mbf with xul 1.8 a blocker"
[13:50] <lool> Okay, I'm sorry my memory is leaky, but I certainly prefer xul 1.9 if it works
[13:50] <asac> and i answered that i would maintain xul 1.8 security wise for a while and we could later upgrade to 1.9
[13:50] <lool> We'll have to iron out why categories don't work anymore
[13:51] <lool> (but launching does)
[13:51] <lool> So .cpp -> xul works and vice-versa
[13:51] <lool> asac: Right, I recall now
[13:53] <asac> yes
[13:53] <asac> .cpp should also work with 1.8
[13:53] <asac> not sure why we dont have that in .0.44
[13:57] <lool> Dunno
[13:58] <lool> Anyway, first lets get it upstream, fix it for other issues (categories), then merge, then backport :)
[14:00] <lool> persia: .svg works with xul 1.9 if you were wondering :)
[14:00] <lool> I'm not too interested in chasing 1.8 to fix svg rendering
[14:05] <pmcgowan> lool, thats good, I will have Bill or Michael contact you later
[14:16] <lool> pmcgowan: asac mailed bspencer; we need to test the upstream patch fully, it seems it has a small regression
[14:17] <pmcgowan> lool, ok, what was regression?
[14:23] <lool> pmcgowan: categories not working
[14:23] <lool> pmcgowan: (marquee menu)
[14:23] <asac> lool: categories like filtering on menu click?
[14:24] <asac> lool: you can set
[14:24] <asac> export NSPR_LOG_MODULES=all:5
[14:24] <asac> and NSPR_LOG_FILE=/tmp/output
[14:24] <asac> and then see what happens if you click the menu
[14:24] <pmcgowan> lool, may not effect our custom build
[14:25] <asac> lool: or use bspencers instructions to get output from the hildon-desktop plugin stdout/err and just set NSPR_LOG_MODULES
[14:25] <asac> but i thin the NSPR_LOG_FILE=... thing is easier to setup and test
[14:35] <lool> asac: Yes, categories are filters for the displayed apps
[14:36] <lool> asac: Sure, will try that
[18:13] <mterry> For UME, where in our scripts is the most appropriate place to set a global environment variable like $BROWSER?
[19:42] <mterry> lool: I can never remember what I'm supposed to do after attaching a debdiff to an LP bug.  This one's yours (LP #191064).  If I hadn't poked you on IRC, would I need to set some further setting so it'd get your notice?
[22:27] <The_Niz> \o
[22:27] <The_Niz> lo
[22:28] <The_Niz> is ubuntu-mobile x86 targeted or also other devices?