All times shown according to UTC.

Time Nick Message
00:30 jupham left #miro-hackers
01:04 willkg left #miro-hackers
01:22 maggie_s left #miro-hackers
01:26 Odysimus joined #miro-hackers
06:35 kcw left #miro-hackers
08:09 z3p left #miro-hackers
08:19 z3p joined #miro-hackers
08:28 maggie_s joined #miro-hackers
10:24 maggie_s left #miro-hackers
11:15 thrillERboy joined #miro-hackers
11:25 thrillERboy left #miro-hackers
12:51 janetPCF joined #miro-hackers
14:10 morganknutson left #miro-hackers
14:14 SNAILBot joined #miro-hackers
14:14 Topic for #miro-hackers is now Miro and MiroCommunity.org developers channel -- Got a bug?  Have the ticket number ready.  Nightlies at http://nightlies.pculture.org/ .  | Channel is logged, "/msg SNAILBot help" for more info.
14:15 zanoi joined #miro-hackers
14:16 morganknutson joined #miro-hackers
14:22 paroneay` joined #miro-hackers
14:25 paroneayea left #miro-hackers
14:27 willkg joined #miro-hackers
14:31 willkg hi!
14:33 willkg left #miro-hackers
14:35 Odysimus left #miro-hackers
14:38 Odysimus joined #miro-hackers
14:41 willkg joined #miro-hackers
14:41 willkg z3p: you use pidgin for irc, right?
14:44 willkg left #miro-hackers
14:45 willkg joined #miro-hackers
14:45 willkg z3p: nevermind...  i think i figured it out.
14:45 CarlFK left #miro-hackers
14:45 CarlFK1 joined #miro-hackers
14:51 willkg left #miro-hackers
14:51 paroneay` left #miro-hackers
14:51 willkg joined #miro-hackers
14:51 paroneayea joined #miro-hackers
15:09 willkg janetPCF: the windows build server hung last night.  i just started it up and i'm running the windows nightly now.
15:11 afk to get a huge cup of coffee.
15:12 one so big, i can drown an elephant.
15:17 janetPCF sounds good willkg
15:40 z3p willkg: I do, but glad you figured out whatever it was :)
15:41 willkg z3p: i was trying to get it to work with irc.  but then i discovered pidgin doesn't support status.net, so i decided the compelling reason to switch wasn't very compelling and i went back to xchat.
15:41 weird that pidgin is missing things that adium has.  i thought they were like sister projects.
15:44 also, the baby is wearing overalls right now.  it's pretty damn cute.  it's hard to sit here and look at miro logs when i could be looking at a baby in overalls.
15:47 z3p willkg: http://status.net/open-source/[…]in-twitter-plugin
15:48 willkg so this screenshot pretty much says it all for me: http://www.honeyplanet.jp/pidg[…]er_screenshot.jpg
15:49 that's just crazy.
15:49 but...  good to know i could do identi.ca stuff with pidgin.
16:25 ajonas joined #miro-hackers
16:49 paulproteus_ ajonas: http://bugzilla.pculture.org/s[…]_bug.cgi?id=14814 describes a limitation of the current implementation -- see comment 8. I just want to make sure you understood that.
16:49 (since you seemed so enthusiastic!)
16:50 ajonas paulproteus_: i understand, it looks good to me
16:51 paulproteus_: people should be aware who their admins are and thus be able to find them in an alphabetized list without needing to sort by role
16:52 paulproteus_ ajonas: Ah hah, I see -- sounds great.
16:54 willkg bendk: i updated miro on osx and windows to use libtorrent 0.15.5.  so we're now using 0.15.5 on all platforms.
16:54 bendk cool
16:55 willkg well, sort of.  now i'm bumping up against some of the torrents-don't-work issues on all platforms.
16:55 :p
16:55 i have to use deluge to download one file so i can test this conversions bug because it won't download in miro in git master.
16:59 afk while this downloads....
17:27 CarlFK1 left #miro-hackers
17:29 paulproteus_ ajonas, z3p: In the OriginalVideo support (checking if the remote server has updated the video, and then updating the video in the MC site), we rely on the remote HTTP server's Last-Modified-Since header to check for updated thumbnails. But I believe that this check causes false positives, so I think it's worth-while to store e.g. the sha1 hash of the image in the database, to double-check that the image was actually changed.
17:30 I'm going to spend a little bit adding that functionality right now (and writing a test for it), but I thought I'd run it by you all first as a sanity-check.
17:30 z3p paulproteus_: sure, that's fine with me
17:31 ajonas paulproteus_: you'll add that to the sandbox/test site, not live, yes?
17:31 z3p paulproteus_: you can probably do the same with the ETag header
17:31 paulproteus_ Oh, screw ETag.
17:31 z3p paulproteus_: and avoid having to download the thumbnail
17:31 paulproteus_ We should just redownload the entire thing, since no one can be trusted to maintain proper HTTP metadata.
17:31 In my curmudgeonly opinion.
17:32 Apache generates broken ETags in the default out-of-the-box config.
17:32 z3p paulproteus_: ?
17:37 paulproteus_ I have read that mod_deflate breaks ETag and that cluster filesystems also break Apache ETag.
17:37 So you can't quite necessarily trust the backend.
17:38 Anyway, whatever.
18:14 z3p: In the OriginalVideo table, we store the remote Last-Modified date. It seems nicer if we store a UTC timestamp of the last time we did a successful *fetch*.
18:15 I guess if I do things that way, I introduce a race condition...
18:16 * paulproteus_ sticks to the old way.
18:16 z3p paulproteus_: if you aren't going to trust the 'last-modified' date, then it doesn't really matter what you store
18:17 willkg paulproteus_: store z3p's birthdate.
18:17 it'd be like an easter egg in the mc code.
18:18 paulproteus_ Hah.
18:18 back in a bit; I'll show you the commit shortly
18:51 z3p: http://git.pculture.org/localt[…]8ec192d2e5f62a26a fwiw
18:54 z3p paulproteus_: other than it maybe not being awesome to load the whole thumbnail into memory, looks fine
19:13 paulproteus_ z3p: Good thinking. I just fixed that with a new function in localtv.util that is called hash_fd that returns a sha1 of all the data from a file object, without slurping it all into RAM.
19:14 z3p paulproteus_: you should probably name it something else, then, esp. since a urllib.HttpResponse isn't a FD :)
19:14 paulproteus_ But it's a file-like!
19:14 z3p paulproteus_: right, but FDs aren't file like; they're numbers
19:14 paulproteus_ Yeah, yeah, it's true.
19:15 I like to pretend in Python we never see C-esque FDs, and call file objects fd.
19:15 I guess my view isn't actually in line with reality, so I accept your suggestion. (-:
20:25 Odysimus left #miro-hackers
20:46 paulproteus_ z3p: er, a question for you: Why do we strip out a bunch of stuff from the ustream.tv embed code, within vidscraper.sites.ustream?
20:46 It seems to break embedding.
20:47 They seem to wrap the <embed> within a <object></object>.
20:48 When I use the full tag as provided by ustream, the embedding works.
20:49 Well, I'll leave that as a question for tomorrow; for now, I'm happy to have diagnosed the problem.
20:51 ajonas: I think that's a wrap for me; you or I can add the above to the bug that you'll file.
20:51 ajonas paulproteus_: ok
20:52 DGMurdockIII joined #miro-hackers
21:01 kcw joined #miro-hackers
21:02 epitron left #miro-hackers
21:03 epitron joined #miro-hackers
21:21 janetPCF left #miro-hackers
21:28 janetPCF joined #miro-hackers
21:36 paulproteus_ z3p: I'm thinking through performance stuff for MC. One real tragedy we have on our hands is that we have a separate process pool for each site.
21:37 I'm wondering (idly, right now) if we can have a master WSGI file that looks at the HTTP Host that it's being called with regard to, and sets DJANGO_SETTINGS_MODULE appropriately.
21:37 That's just a random thought (that maybe you'll see tomorrow).
21:44 morganknutson left #miro-hackers
21:56 DGMurdockIII left #miro-hackers
21:58 morganknutson joined #miro-hackers
22:27 janetPCF1 joined #miro-hackers
22:29 janetPCF left #miro-hackers
22:35 janetPCF joined #miro-hackers
22:37 janetPCF1 left #miro-hackers
22:42 willkg left #miro-hackers
22:50 janetPCF left #miro-hackers
23:18 maggie_s joined #miro-hackers
23:26 kcw left #miro-hackers
23:31 z3p paulproteus_: I had the same thought, actually
23:31 paulproteus_: but as I recall, it ended up not working well enough to use

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