Web   ·   Wiki   ·   Activities   ·   Blog   ·   Lists   ·   Chat   ·   Meeting   ·   Bugs   ·   Git   ·   Translate   ·   Archive   ·   People   ·   Donate

#sugar-meeting, 2009-08-11

Index | Today     Channels | Search | Join

All times shown according to UTC.

Time Nick Message
10:14 bemasc #topic administrative
10:14 "Pencils down", i.e. the end of the official GSoC period is approximately one week from today, correct?
10:14 aa right
10:15 bemasc So we will have at least one more meeting.  This also means that mentors will need to start thinking about final reviews, I presume.
10:16 Any other administrative business?
10:16 silbe what do we submit to google?
10:16 bemasc It seems we have given up on t-shirts and local deployments, which is fine by me.
10:16 lucian silbe: tarballs of all our commits. git fast-export essentially
10:17 subzero diff files or the whole file?
10:17 silbe so each single commit, not a tarball of the "finished" product?
10:18 lucian silbe: yes
10:18 bemasc In fact, Google lists today as a "suggested pencils-down date".
10:18 i.e. they suggest today as a feature freeze.
10:19 iwikiwi That means im still free to optimize my code?
10:19 bemasc You can do whatever you want (talk to your mentor).
10:19 I'm referring to the suggestions on this page: http://socghop.appspot.com/doc[…]009/faqs#timeline
10:19 iwikiwi Oh okay
10:20 aa: believe we will be needing volunteers to test code on the XS
10:20 bemasc ok, next topic is progress report from subzero
10:21 #topic subzero
10:21 subzero I'm working with a new version of karma: http://git.sugarlabs.org/proje[…]r/js/karma-0.2.js
10:22 I have fixed some little bugs
10:22 and found the reason of the security problem when calling getimagedata
10:23 it's about the "origin" (file vs http) of the element
10:24 lucian subzero: XSS is horrible, i know
10:26 bemasc subzero: are you done?
10:26 hmm.
10:26 #topic iwikiwi
10:27 subzero1 sorry, do u get my msgs?
10:27 aa iwikiwi: still there?
10:27 bemasc #topic subzero
10:27 iwikiwi aa: yeah
10:27 bemasc subzero1: ended at < subzero> it's about the "origin" (file vs http) of the element
10:28 subzero1 also, I have added KGroup to karma-0.2js that supports z-index and I'm using 1 single canvas to render the objects
10:28 bemasc: thanks
10:28 about the image-rendering , it's available with ff 3.6alpha, I'm playing with it
10:29 but, obviously it is not availabe with the current Browse or surf :(
10:29 that's all
10:30 bemasc ok
10:30 #topic iwikiwi
10:30 iwikiwi Apart from fixing the authentication stuff with a colab session with aa. ( which involves creating a cookie on the local machine with the pub key and and using a POST to again connect to martins auth plugin for auhthenticating the user) I've also managed to write a function which gets me the instances of the print module inscribed courses. Right now I am writing a sugar interface to select which course to send a pdf from moodle.
10:31 from sugar*
10:31 With that at least a complete working model will be done
10:31 bemasc iwikiwi: have you released activity bundles for the Print activity?
10:32 iwikiwi bemasc: nope, I'm testing them myself though. But I will today later, since everything will be done
10:33 bemasc ok
10:33 wonderful
10:33 aa iwikiwi: I want to work on documentation next
10:33 bemasc I think it's important for you to release activity bundles, so that people can easily test it and make sure it works.
10:33 iwikiwi point taken. I will do that tonight.
10:34 aa: yeah, but what about shifting to http transport for the file transfer?
10:34 believe base64 inserts a massive overhead
10:34 for larger files
10:34 aa iwikiwi: um, how are we not using http now?
10:34 bemasc or more importantly, "Make sure that it stays working."  In 0.86, Print should still work.
10:36 iwikiwi aa: we are, but doesnt doing a direct post use some other algorithm?
10:36 bemasc: yep, agreed.
10:36 s/use/using
10:37 aa iwikiwi: lets talk about this in #sugar
10:37 iwikiwi okay
10:37 any questions? bemasc, silbe?
10:37 bemasc iwikiwi: base64 is a significant overhead, but it's much less than, say, the CPU time needed to actually process the PDF.  Don't let it hold up a release.
10:38 silbe iwikiwi: not for now - busy with my own stuff, sorry.
10:38 iwikiwi bemasc: hmm, so basically its less than what cups-pdf utilizes.
10:39 bemasc iwikiwi: both producing the PDF, and converting the PDF into a bitmap for printing (all printers are bitmaps at the hardware level) require more CPU power than the base64 conversion).  It wastes about 50% in bandwidth too.
10:39 It's not great, but it's not fatal.
10:40 #topic silbe
10:40 silbe did another round of review with Tomeu, waiting for the next one
10:40 iwikiwi bemasc: Ah, I can see.
10:40 silbe worked on a test suite for the unaltered datastore
10:41 revived some old testing code and modified it to talk via DBus instead of directly to the classes
10:42 get some basic tests working fine, now need to add more
10:42 *got
10:42 bemasc silbe: you have a very big project. what would you like to accomplish by next week?
10:43 silbe bemasc: good question. would like to see my code enter mainline, but that doesn't seem realistic.
10:43 bemasc silbe: do you have a demonstration of an activity using your new API?
10:44 maybe you could build enthusiasm for getting your code into mainline by making a screencast showing a compelling use case?
10:44 silbe i'll continue working on the test suite, adding one for the new API as well
10:45 bemasc ok
10:45 silbe bemasc: as i tried hard to minize changes for existing activities, there's no demo activity using the new API
10:45 *minimize
10:46 but i can try coming up with some screencast...
10:47 bemasc silbe: I don't think it would be too hard.  I think people would be interested to see what you have in mind, and it might help get more people looking at the code.
10:47 any more for silbe?
10:47 aa silbe: a screencast would be awesome ;)
10:48 bemasc #topic lucian
10:48 silbe ok, will definitely try to do one. i have a feeling it's going to take quite some time, though... :-/
10:48 lucian i've been polishing things and playing around with websites in Browse and SSBs
10:49 GMail works great, even the offline function
10:49 bemasc lucian: where is the offline state kept?
10:49 lucian i've decided to include a whitelist permissions.db for Gears, so that most gears websites work
10:50 bemasc: Gears state? in /data/ right now. it's unrealistic to put it in the journal
10:50 it grows very large with just my gmail inbox
10:50 bemasc lucian: really? you can't just tar it up into a journal entry?
10:50 lucian it would take 15secs to zip that and put it in the journal
10:50 same with tar, i used zip without compression
10:51 it gets HUGE very fast
10:51 bemasc ok
10:51 lucian it's not much of a problem, though
10:51 bemasc can you set a local cache maximum size?
10:51 lucian Gears web sites are designed to handle sudden death
10:51 bemasc I'd hate to mysteriously run out of disk space...
10:51 lucian bemasc: gears does that by itself, but i'm not sure how large the limit is, i'll have to check
10:52 bemasc ok
10:52 I think you might want to lower the limit a bit
10:52 lucian the only problem with having Gears state in /data/ is that it breaks continuity when version control works
10:52 bemasc I wouldn't say that's the only problem
10:53 The great thing about keeping browser state in the journal is that it acts like a profile mechanism.
10:53 Specifically, it provides information isolation.  It even allows something akin to FF3.5's "anonymous browsing" mode; just start a new session, and then delete its journal entry.
10:53 lucian bemasc: other similar state is already stored in /data/, so at least in Browse's current state, it's not evident
10:53 bemasc yeah
10:54 this is a problem in Browse generally.
10:54 lucian i'm not sure i'd like all state do be in the Journal, but that's another discussion
10:54 silbe lucian: would be great if you tried not to make things worse, though
10:54 lucian i've also worked a bit more on 'save complete page'
10:55 silbe: Gears is just a temporary solution until everyone moves to html5
10:55 silbe lucian: and with html5 it gets saved into the journal?
10:56 lucian silbe: it gets saved along with other browser data, wherever that is
10:56 walterbender well, html5 isn't going to be ready by next week :)
10:56 we need to keep in the scope of GSOC
10:56 lucian in any case, it's not too hard to change it to save stuff in the journal
10:56 bemasc lucian: what the mime type of saved pages?
10:57 lucian bemasc: application/zip
10:57 bemasc lucian: in other words, Browse will now open any application/zip
10:57 silbe lucian: and where does it save the "other browser data"?
10:57 lucian read_file looks for index.html or other html files in files with zip
10:57 silbe: right now, in /data/
10:57 bemasc lucian: ok.  Are you familiar with the .xol bundle system?
10:57 lucian bemasc: not very, but isn't that deprecated?
10:58 silbe lucian: ok, so you're not talking about bookmarks (which get saved to the journal), but about preferences
10:59 lucian silbe: preferences, location bar, etc
10:59 bemasc lucian: I'm not sure whether .xol is deprecated... but it's just a zip file with an index.html
10:59 lucian: in fact, so is every .xo bundle
11:00 lucian i know they're all zip :)
11:00 so i've just reinvented xol?
11:00 bemasc what I'm saying is, you've just built us a very convenient, universal zip explorer
11:00 which has a lot of applications
11:01 so I hope you'll think about some of the other interesting uses for it
11:01 lucian bemasc: i guess browse could do a listdir if there's no .html inside
11:01 bemasc that would be very helpful
11:02 walterbender we should quickly sum up what Lucian should do to wrap up GSOC
11:02 and then talk about next steps...
11:03 lucian is not going to be allowed to drop out after the summer :)
11:03 lucian walterbender: i should try to make small patches against Browse mainline with my added features
11:03 walterbender: heh
11:03 bemasc lucian: I think you should replace Browse mainline.
11:03 lucian bemasc: it's grown some features since i forked it
11:03 walterbender lucian: yes. I agree. Getting your work into mainline is a good goal for wrapping up GSOC
11:03 bemasc ok
11:03 yes
11:03 walterbender even though GSOC doesn't expect commits...
11:04 lucian walterbender: yes it does, annoyingly :) they want tarballs of all commits
11:04 bemasc weirdly, not until september
11:04 ok
11:04 lucian walterbender: since i commit often, i already have loads of commits
11:04 walterbender lucian: I mean, they don't expect merges
11:04 lucian oh, right
11:05 i'm also waiting on some design feedback for 'save complete page'
11:05 bemasc lucian: I think that should be the default "keep" behavior.
11:05 walterbender +1
11:05 lucian bemasc: i've thought of that too
11:06 besides that, i haven't worked on webcollab any more
11:06 it's just a proof of concept right now
11:06 bemasc that's fine.
11:06 lucian anyway, it proved the concept :)
11:07 bemasc I think it's time to wrap up
11:07 #topic bemasc
11:07 lucian sure, i'm done anyway
11:07 bemasc http://lists.sugarlabs.org/arc[…]ugust/000067.html
11:07 Pippy-35 includes live shared editing of Python code, syntax highlighting and all.
11:08 It should even support automatic merge of changes made offline, though that's a bit... undertested.
11:09 I can't think of any new software features that could be added easily, so all that's left is documentation and cleanup.
11:09 That's all for my report.
11:10 any final remarks? walterbender?
11:10 silbe bemasc: congratulations!
11:10 bemasc thanks silbe
11:10 oh, it also supports distributed Undo and Redo, which is pretty cool because I definitely made no attempt to implement it.  I got it for free from gtksourceview.
11:11 iwikiwi bemasc++
11:11 silbe bemasc: wow!
11:11 walterbender thanks for running the meeting...
11:11 lucian bemasc: congrats!
11:11 bemasc thanks all
11:11 #endmeeting

Index | Today     Channels | Search | Join

Powered by ilbot/Modified.