All times shown according to UTC.

Time Nick Message
01:16 DGMurdockIII left #miro-hackers
01:16 maggie_s left #miro-hackers
01:28 willkg left #miro-hackers
02:52 Olena-QA joined #miro-hackers
02:55 Olena-QA left #miro-hackers
04:21 Odysimus left #miro-hackers
04:24 Odysimus joined #miro-hackers
04:32 CarlFK joined #miro-hackers
06:35 CarlFK left #miro-hackers
06:47 JasonWoof Hi all, just updated to latest git master. Had to do this: http://paste.debian.net/108797/ to get it to build/run on debian unstable
06:48 glee JasonWoof: I'll go fix it up
06:48 JasonWoof glee: awesome :)
06:49 I love the new linker errors that tell you what library contains the symbol
06:50 paulproteus (-:
06:52 JasonWoof and now the interface has gone non-responsive, while the memory usage slowly approaches 1GB
06:52 CPU in the 7-15% range mostly
06:53 ok, killing it now at 1.1GB
06:54 glee I've seen a handful of problems with excessive memory usage, it'd be fantastic to know what's actually wrong so we can fix it
06:58 JasonWoof: [master d2c7dbb]
07:01 JasonWoof glee: what repo? I'm not seeing that revision at git://git.participatoryculture.org/miro/
07:01 glee JasonWoof: You are too fast!  It just got pushed, it should be there now
07:02 JasonWoof ahh, now I see it.
07:03 glee: you're fast too! Most errors I report don't get fixed within minutes :)
07:03 glee :)
07:07 JasonWoof wish I could provide more details about the hang, but I didn't see anything useful-looking in the terminal. Now miro is running fine (with your new patch, instead of mine) and using ~300MB of ram
07:10 glee Might be a good idea to try out some leak debugging in Python with gc but that's probably a lot of work
07:28 JasonWoof glee: well, I'd need to be able to reproduce it. it only happened on that first run that upgraded my database
07:29 glee JasonWoof: We haven't been able to reproduce the ones in the bug database either, which kind of sucks :(
07:36 JasonWoof is it still possible to see info about bittorrent traffic (particularly the current upload rate and total uploaded)
08:02 odenthaa joined #miro-hackers
08:02 odenthaa left #miro-hackers
11:26 willkg joined #miro-hackers
11:44 willkg hello!
12:53 deardiary: discovered some problems with running unittests, wrote up bugs, fixed nightlies_check script.
12:53 deardiary: running pyflakes on portable code.  found some bugs.
13:01 JasonWoof mmm, I gotta try again on profiling youtube downloads
13:02 I realized after trying before that it's the MiroDownloader.py process that goes bad
13:06 willkg goes bad?
13:44 CarlFK joined #miro-hackers
13:55 mick_laptop joined #miro-hackers
14:16 willkg mick_laptop: hi!
14:35 z3p good morning
14:37 mick_laptop hi willkg ;)
14:45 JasonWoof willkg: my MiroDownloader.py process slowly (over about 5-10 minutes) approaches 100%cpu when downloading a long youtube video (eg a google tech talk)
14:45 bendk joined #miro-hackers
14:45 bendk hey all
14:45 willkg: so 16431 seems like an issue with the build box not having the right dependencies to actually run miro
14:45 JasonWoof willkg: it does successfully download, it's just no fun to have that cpu drain for 30-60 minutes
14:46 bendk I think the best way to fix at this point would be to just install them there.  It seems like the main issue is that our current archetecture requires you to import widgetset if you want anything inside it, and changing that would be a pretty huge deal
14:46 willkg bendk: really?  are you sure it's not a problem that because it's running from cron, there's no display?  i thought we agreed that unit tests wouldn't be pulling in stuff from widget-land.
14:46 bendk actually, though, let me take 5 minutes to see if I can come up with a better shorterm solution
14:47 so that test in particular is trying to test the InfoListModel, which comes from widgetland, but doesn't really need any graphical stuff
14:47 I do remember saying unittests don't need widget stuff, but it seems like that isn't working anymore
14:47 willkg maybe it makes sense to add a "--nodisplay" flag and some additional skip stuff?
14:48 bendk possibly, but I would like to run that test if we could
14:48 * willkg nods.
14:48 bendk yeah, you're totally right about it being about not having a display
14:51 JasonWoof sounds like you might have to separate out unit tests (tests on very small pieces of code with fake dependencies usually) from the bigger tests that take a while and/or must be run on your computer
14:51 willkg JasonWoof: we've been building infrastructure for that.
14:52 bendk willkg: where is the documentation to run the nightlies on the build box again?
14:52 sorry I keep forgetting
14:52 JasonWoof willkg: cool
14:52 bendk not the buildbox one, but the instructions to find the script that disables dbus
14:52 willkg bendk: i usually just do "crontab -l".
14:52 bendk: that tells you what's running at night.
14:53 zanoi morning!
14:53 willkg bendk: we can probably turn the dbus disabling into an alias or a script or something on the box.  definitely feel free to make your life easier when doing things you're doing a lot.
14:54 JasonWoof: yeah, it's been a long process.  so far we have support for skipping tests depending on what platform you're on.  i want to extend that to group tests and test suites into other dependency groupings.
14:54 JasonWoof: which would cover network connectivity, display, ...
14:54 bendk I'm thinking I might check those commands into the source tree
14:54 willkg bendk: that's fine, too.
14:56 bendk cool, seems like that is there now
14:57 willkg z3p: i think the instructions for finding the usb name on ubuntu need to be clearer.  maybe it's better to have a link back to the DeviceSupport page where we can have clearer instructions and screenshots?
14:57 z3p: if you want, i can write that up on DeviceSupport and then if it's ok with you, then you can add the link to the form.
14:59 bendk okay, so I think I have a fix that lets us things work again.  Ideally, I'd consider breaking up widgetset into widgets and "other stuff", so that we could import InfoListModel without importing our widgets
14:59 but I'm probably going to leave that for another day
14:59 z3p willkg: sounds good
14:59 * willkg nods.
15:00 willkg z3p: ok.  i'll work on that now.
15:00 bendk I'm +0 on merging the extensions manual
15:03 * willkg nods.
15:03 willkg it saves me some work.  so i figured i'd mention it.
15:11 bendk willkg: I just made a script that messes with DBus to run the unittests, but i didn't "merge" it into your nightly script
15:11 tell me if you think I should
15:12 willkg bendk: where's the script?
15:12 bendk tv/linux/run-unittests-no-display.sh
15:12 which actually should be "no-dbus"
15:12 I'll fix that name
15:13 willkg bendk: that also doesn't handle other command line unittest flags.
15:14 bendk: but...  it's fine.  you can merge it into the nightly script if you want.
15:14 but i don't think it's necessary.
15:17 bendk hmm, I think that unittests don't currently support options at all
15:17 if I run ./run.sh --unittest -v, I don't get verbose output
15:18 willkg mmm...  not sure why -v isn't working.  but --failfast and passing in specific test suites to run works.
15:18 i'll look into -v.  i know i had it working a month or two ago.
15:19 bendk I'll fix that too
15:22 okay, script is better now.  I'm just going to leave the nightly script alone
15:23 willkg ok.
15:23 thanks for finding and fixing -v.  it might have gotten screwed up when i implemented --failfast.
15:24 zanoi where do we register which mimetypes miro is associated with?
15:25 willkg bendk: i'm going to rename run-unittests-no-dbus.sh to "test.sh" which is what we use in the other platforms.
15:25 bendk: is that ok?
15:26 bendk no problem
15:36 willkg bendk: i threw bug #16429 at you.  i fiddled with it for a while and then gave up.
15:37 bendk: one of the issues can be fixed by having HideableWidget pass the get_size_request to its hidden child.  the other one i have no idea about.
15:37 bendk: if you need more information on that, let me know.  it's related to the pref panel fixes i did a week or two ago.
15:37 bendk heh, I'm way ahead of you :)
15:37 well 1 step that is
15:38 willkg yay!  :)
15:39 on that note, i'm going to go put on pants.
15:42 bendk: the set_size_request is a great idea!  i should have thought of that.  bah.
15:42 afk for pants.  for realz.
15:42 glee left #miro-hackers
15:45 z3p left #miro-hackers
15:46 z3p joined #miro-hackers
15:56 zanoi willkg: do you know if mimetype association is done on the packaging level (as opposed to the source code level)?
15:59 glee joined #miro-hackers
16:00 z3p zanoi: on windows, it's done as part of install
16:00 zanoi: I don't know about other platforms
16:01 zanoi z3p: ok
16:03 ajonas joined #miro-hackers
16:42 willkg zanoi: when you say "mimetype association" are you asking about the miro browser or the operating system's browser?
16:43 zanoi operating system browser
16:43 willkg zanoi: on linux, there's a miro.desktop file which has a MimeType line.
16:44 zanoi d'oh! right
16:44 willkg zanoi: not sure how windows works.  i know z3p just said it was part of th einstall, but i think that's file types and not mimetypes.
16:44 and i have no idea how it works on osx.
16:45 glee: ^^^  any idea how miro associates with mimetypes on osx?
16:46 zanoi willkg: thanks
16:47 janetPCF joined #miro-hackers
16:47 glee Not sure with the browser per se off hand, sorry.  Filetypes is Info.plist.
16:48 willkg zanoi: so i think the answer is that miro associates with filetypes on windows and osx and mimetypes on linux.
16:49 zanoi hmm...but a magnet doesnt have a filetype
16:50 glee zanoi: Let me pastbin you
16:50 Sorry, I didn't know your context, now I see what you are trying to do.  :)
16:51 zanoi the Info.plist refers to "application/x-bittorrent" so it's probably in there
16:51 glee zanoi: http://pastebin.com/yKTxpapQ
16:51 willkg janetPCF: when you download a media file with a browser on windows, does it get opened in miro?  have you seen it ever do that sort of thing?
16:52 glee Add that to the Info.plist
16:52 zanoi glee: great, thanks
16:52 janetPCF if you have the file associations setup
16:53 glee zanoi: Ah, I missed a line at the top, prepend <key>CFBundleURLTypes</key>
16:53 zanoi sure
17:07 janetPCF1 joined #miro-hackers
17:08 zanoi willkg: i think we should probably associate with all ogg related mime types
17:08 and file extensions
17:08 willkg zanoi: write up a bug for it.
17:08 zanoi will do
17:10 janetPCF left #miro-hackers
17:17 ajonas left #miro-hackers
17:35 willkg z3p: so...  i'm trying to fill out the directions for finding the device information on Windows and i don't see it here either.
17:35 z3p: i'm on Windows Vista and looking at the Device Manager.  i see a list of USB devices, but i don't see USB names, vendor ids, or product ids.
17:35 z3p willkg: let me pull up my VM
17:36 willkg z3p: it's probably the case that it's different between Windows XP, Vista and 7.
17:37 z3p willkg: so for me, the name shows up under Disk Drives
17:37 willkg: and the VID/PID show up under USB Controllers > USB Mass Storage > Device Instance ID
17:38 willkg man...  i never would have figured that out from the current directions.
17:42 z3p willkg: yeah...
17:42 willkg: it's tough when there's only a tiny text field
17:43 * willkg nods.
17:43 CarlFK left #miro-hackers
18:06 willkg afk a bit to shower, get lunch, ...
18:12 janetPCF joined #miro-hackers
18:13 janetPCF1 left #miro-hackers
18:46 paulproteus z3p: Sanity-check: As far as I can tell, the vimeo embed API generates invalid nonsense for the http://www.ustream.tv/recorded/8272302 video. So I'm going to instead use an iframe that wraps a different part of their API. (This is a vidscraper change.)
18:47 <iframe src="http://www.ustream.tv/flash/video/8272302" width="320" height="260" /> # will be the new embed_code() return value
18:47 z3p paulproteus: your statement is confusing
18:47 paulproteus Yeah, I can see that. Sorry, half paying attention at SCALE.
18:47 z3p do you mean ustream for everything?
18:47 paulproteus Er, and now I've confused myself.
18:47 tty in a sec
18:47 uwe_ left #miro-hackers
18:48 paulproteus Okay, rehi. (-:
18:50 uwe_ joined #miro-hackers
18:52 paulproteus z3p: So here's how I see things. 1. The embed HTML source at http://www.ustream.tv/recorded/8272302 works great.
18:52 2. None of the embed source at http://api.ustream.tv/html/video/8272302/getInfo/ works.
18:52 z3p awesome
18:52 paulproteus 3. I hate people's APIs when they're worse than the real site.
18:53 4. There's hope -- there's a embedTagSourceUrl that the API gives us, and if we wrap in an iframe, we can render the videos just fine.
18:53 z3p paulproteus: sounds good to me
18:53 paulproteus Great.
19:02 CarlFK joined #miro-hackers
19:58 bendk left #miro-hackers
20:45 willkg deardiary: working on instructions for finding the information to fill out the form for adding new device support.
21:21 janetPCF left #miro-hackers
21:30 janetPCF joined #miro-hackers
21:42 Sergeiqa joined #miro-hackers
22:17 bendk joined #miro-hackers
22:17 janetPCF left #miro-hackers
22:50 morganknutson joined #miro-hackers

← Previous day | Index | Server Index | Channel Index | Today | Next day → | Atom Feed | Search | Google Search | Plain-Text | plain, newest first