All times shown according to UTC.

Time Nick Message
00:04 _epitron_ is now known as epitron
00:06 zanoi ajonas, paulproteus: All the URLs I add on Mirocommunity seem to be importing data automagically
00:06 except for TED urls
00:06 paulproteus We probably don't have support for TED in vidscraper.
00:06 zanoi am I doing something wrong is that not supported yet?
00:06 paulproteus I'm not sure, though.
00:06 ajonas zanoi: youtube, blip, vimeo, ustream and others add automatically
00:06 zanoi ok, thx
00:06 paulproteus I mean, basically, I'm pretty sure that's the reason, unless there's some insidiously awesome magic I don't know about. (-:
00:06 ajonas zanoi: i think fora, too, i think that's it
00:06 zanoi yes fora did
00:07 since Fora worked I was wondering that TED didn't :-)
00:09 willkg i'm done for the night.  see you all tomorrow.
00:09 willkg left #miro-hackers
00:14 maggie_s joined #miro-hackers
00:21 ajonas left #miro-hackers
00:57 maggie_s left #miro-hackers
02:42 xfxf joined #miro-hackers
03:54 Odysimus left #miro-hackers
04:04 kcw left #miro-hackers
04:30 epitron left #miro-hackers
04:30 ChanServ left #miro-hackers
04:30 janetPCF left #miro-hackers
04:30 mattjohnson left #miro-hackers
04:30 jamessan left #miro-hackers
04:30 janetPCF joined #miro-hackers
04:30 mattjohnson joined #miro-hackers
04:30 epitron joined #miro-hackers
04:30 ChanServ joined #miro-hackers
04:34 jamessan joined #miro-hackers
05:45 Odysimus joined #miro-hackers
08:10 odenthaa joined #miro-hackers
08:10 odenthaa left #miro-hackers
09:01 Dawnrell joined #miro-hackers
09:18 Dawnrell left #miro-hackers
10:26 Odysimus left #miro-hackers
10:32 Odysimus joined #miro-hackers
11:54 willkg joined #miro-hackers
12:05 janet_ left #miro-hackers
12:41 willkg deardiary: removed a bunch of spam from the wiki.
12:41 hello everyone.
12:54 deardiary: synced translations to Miro-3.5 branch for the last time.
13:19 janetPCF left #miro-hackers
13:25 janetPCF joined #miro-hackers
13:30 z3p good morning
13:31 paulproteus Morning, z3p!
13:33 z3p paulproteus: how's the paypal nonsense coming along?
13:34 paulproteus z3p: Pretty nicely, actually. I'll have more news for you (all) by noonish, but I think all the fighting is over, and now it's a simple matter of programming.
13:36 z3p paulproteus: good to hear
13:56 willkg afk for coffee and some food.
14:01 paulproteus deardiary: writing up weekly notes, then integrating paypal some more.
14:02 willkg janetPCF: how much coverage do we have with the sikuli tests now?
14:02 janetPCF: where "coverage" can be defined however you like.
14:20 zanoi willkg: I ran Miro in danish and it had some pretty stupid mistakes
14:20 should I maybe try to go through miro in all the languages I know and see if the translations are correct?
14:20 willkg zanoi: you can update/fix the translations on launchpad: https://translations.launchpad.net/democracy
14:20 yeah, if you could, that'd be great.  but ...  i'd wait until strings are stabilized.
14:21 zanoi willkg: ok, sounds good
14:21 willkg theoretically, the march 1st code freeze will be a good time for a strings freeze, too.
14:22 zanoi do you mind if I write myself a bug report so I won't forget?
14:22 willkg but historically, we kind of suck at sticking to strings freezes.
14:22 we tend to close translations-related bugs unless they involve a technical problem.
14:22 zanoi ah, ok
14:22 willkg so if you're going to do that, make sure it has some text stating not to close the bug.
14:23 zanoi it's fine, i'll just add it to my personal todo list instead
14:26 willkg mmm...  i can't remember where i left off yesterday.  how irritating.
14:27 z3p deardiary: working on #16049
14:44 janetPCF willkg:  I haven't counted it yet
14:44 * willkg nods.
14:44 janetPCF willkg:  at this point, I have more tests in more different subgroups than what i had before
14:44 but there are some subgroups that I haven't finished yet
14:45 some of the tests relied on the more info fields that are not currently present
14:46 and right now - I have to go through and cleap of some details so they don't fail
14:46 precision timing and all
14:46 * willkg nods.
14:47 willkg how useful are they now?
14:47 janetPCF they are useful to me, because things that fail are the test fault - and not crashes -
14:47 but they wouldn't be useful to anyone else yet
14:47 * willkg nods.
14:48 janetPCF I'm still waiting on rc2 from sikuli that keeps being 1 day away
14:48 willkg heh.  i know that tune.  :)
14:48 janetPCF then I can finish up some details and images that I'll need for windows and osx
14:48 willkg i can play it in four-part harmony, myself.
14:48 janetPCF windows and linux
14:48 then they will be more more useful
14:49 z3p paulproteus: re #16177, is there any reason you can't put the CSRF token directly in the thumbnail template?
14:49 willkg i wonder how much i spike the DemocracyNow numbers when i'm testing torrent issues.
14:53 janetPCF left #miro-hackers
14:53 willkg sweet.  i think i got most of this fastresume overhaul working.
14:56 paulproteus z3p: I... guess. It just seems so evil to put it there, outside of a <form> tag.
14:56 But yeah, that would do the trick too.
14:56 I think.
14:56 I'd have to double-check. z3p, one big problem is that the form is being generated by a template tag, not the normal RequestContext process.
14:57 zanoi willkg: would you have time at some point today to walk me through where to add the magnet links?
14:57 paulproteus Yeah -- and the template tag doesn't have access to the CSRF data unless it has access to the request.
14:57 willkg zanoi: doing that requires me to go figure out where to add magnet uri handling code.  that's the thing i haven't had time to do, yet.
14:57 zanoi: so i don't have the knowledge you seek.
14:58 paulproteus z3p: I don't know if you have any expertise that could help, but maybe: http://bugzilla.pculture.org/s[…]_bug.cgi?id=13607 is about a bug where a page works fine in Fx and hangs Chromium, and I can't even get Chromium to not-hang for long enough so I can run its profiler and figure out what of my JS is breaking it / causing an infinite loop.
14:58 willkg zanoi: generally, it's all the places that miro looks at a uri and figures out what to do with it.
14:58 zanoi: this covers some/most of them, i think: http://manual.getmiro.com/downloading.html
14:59 zanoi willkg: ok, i'll have a look around the code
15:00 willkg zanoi: also, if your comment #8 in bug 10381 is there hoping to solicit input from other people, then you should phrase your questions as questions.  otherwise, people don't know what it is you're asking of them because you're not asking anything.
15:01 z3p paulproteus: can you add some alert()s? they should stop the interpreter and if you get a ton of them you'll know where your infinite loop is
15:01 paulproteus z3p: Oh, that's reasonable.
15:01 Thanks, that makes good sense.
15:02 zanoi willkg: well, i think i know what they meant, so I only need them to comment if they disagree with what I think
15:02 willkg ok.
15:03 just know that almost everyone in the cc: list except elmargol is busy and not reading everything they're getting in email carefully.
15:05 zanoi sure
15:06 oh man my internet is slow, it will be an "interesting" dev talk
15:09 willkg i forget if i got coffee or not earlier.  assuming that i didn't, i hereby embark on this ambitious quest.
15:12 zanoi willkg: thanks for your email :)
15:14 Dawnrell joined #miro-hackers
15:15 janetPCF joined #miro-hackers
15:18 * zanoi swears at his internet
15:18 paulproteus Now, now, zanoi! Patience.
15:19 zanoi patience is overrated!!
15:19 ajonas joined #miro-hackers
15:19 zanoi i can't even load websites right now
15:20 paulproteus Then, like, call your ISP?
15:21 Start using the wondershaper, if you think the problem might be related to other things you're doing.
15:21 Those are my two main recourses, and they've served me reasonably well. Morning, ajonas.
15:21 ajonas morning all
15:23 paulproteus FWIW, I'm working on PayPal and transitions between different tiers.
15:27 kcw joined #miro-hackers
15:28 ajonas paulproteus: cool beans
15:31 paulproteus ajonas: In no-enforce mode, do you want us to skip the payment validation also?
15:31 ajonas paulproteus: seems likely - what's that, exactly?
15:31 paulproteus In particular: if you are in e.g. plus and want to upgrade to 'max' but your site is in no-enforce mode, do I send you to paypal?
15:32 zanoi paulproteus: yes i'm pretty sure it's the ISP
15:32 paulproteus no-enforce mode is http://bugzilla.pculture.org/s[…]_bug.cgi?id=16145 ajonas
15:33 ajonas paulproteus: here's the problem
15:33 paulproteus: for those in the "waiting period" the answer would be yes - because they should be on "no enforce" on basic, so that they can upgrade, and then we want them to go to paypal
15:33 paulproteus: but for subsidized sites, they shouldn't go to paypal
15:33 paulproteus: thoughts?
15:34 paulproteus 1. We could set subsidized sites to 'max'?
15:34 But maybe we don't want to do that for some reason.
15:34 2. If subsidized sites really want to change their site tier, isn't that kind of crazy?
15:34 ajonas paulproteus: i think that's a good solution
15:34 paulproteus This is why I think there should be a message at the top of the Upgrade screen -- if you're on no-enforce mode, the screen should say why.
15:35 ajonas so then, yes, they should get sent to paypal
15:35 paulproteus bd
15:35 Life is easy.
15:35 ajonas paulproteus: they'll have had an email explaining why
15:35 paulproteus Good enough for me, I guess!
15:35 ajonas paulproteus: and this is a temporary, one time concern ,really
15:35 just for the transition
15:35 after that it'll just be subsidized sites and as you said, they can be set to max
15:36 paulproteus Ah hah, true enough.
15:37 bendk joined #miro-hackers
15:40 Dawnrell left #miro-hackers
16:01 * paulproteus tests his paypal redirects in a web browser.
16:36 ajonas paulproteus: for http://bugzilla.pculture.org/s[…]_bug.cgi?id=16038, i don't want the poor user stuck on the page while the site adjusts for 30 min. can we just have a pop-up that tells them what will happen and then have that going on in the background, or something?
16:36 paulproteus Yeah, that's fine with me.
16:37 ajonas i apparently love pop-ups, i'm always wanting more of them. i suspect that is bad.
16:37 paulproteus (-:
16:40 ajonas paulproteus: when a ticket is waiting on a patch from me, can you re-assign it to me? then i'll know what i need to add and can re-assign when finished
16:40 paulproteus I can do that going forward; sounds grea.t
16:40 ajonas ty
16:46 paulproteus: for the about page, do i just need to make a patch for the blue theme (4), or for all of them?
16:47 paulproteus I think just blue (4), since I think the rest don't change it.
16:47 ajonas ok
16:47 paulproteus If possible, we should use a template fragement file, honestly.
16:47 ajonas what's that?
16:48 paulproteus It's a way to store a fragment (not fragement; that was a typo) of HTML in a file, and have other files include it.
16:48 If it's in more than just blue and the stock template, then I think we should do that.
16:48 ajonas ... ok ? if you think we should do it, go for it.
16:49 paulproteus: i mean, if that's the best way, it's cool with me - so i should not file a patch?
16:49 paulproteus ajonas: Well, if it's just the blue theme, then make a patch.
16:50 If you discover that we need to change other templates, too, then leave a note kicking it back to me.
16:50 ajonas paulproteus: i think the issue is that i had already made the change to the other template
16:50 paulproteus So I'm assigning some research to you, too, basically (-:
16:50 Yeah -- that's fine. It's a question of how many changes need to be made inside localtv-themes.
16:50 If it's more than one file in localtv-themes, we should do it my template fragment way.
16:50 ajonas paulproteus: this is for future reference?
16:51 paulproteus ajonas: Basically, make your patch, but if you discover that there's more than one file in the localtv-themes repo that ought to be changed (e.g., because not just the blue theme has this problem) then I will actually reject your patch and do it my way.
16:51 I actually think that it'll be just the blue theme, so you patching it will be fine.
16:51 I hope that makes sense?
16:52 I'm concerned that if we need to be adjusting *lots* of themes, then we should do it in a generic way. But if it's not lots of themes it's okay to be non-generic.
16:54 ajonas paulproteus: i think in general doing things in the generic way is preferable. it sounds like in this case it would have been better to start with if we'd done it that way rather than having me make a patch which was checked in and now making another patch. i'm not thrilled with the idea of doing the work to make the patches and then having them rejected - if that's not the way to do it, we should figure that out up front. so it soun
16:54 like in this case the ship has sailed and i'll go ahead and do this one, but in the future it seems like it won't actually make sense for me to do front end facing patches, because those are going to need to be cross-theme. sound right?
16:54 paulproteus It's a matter of *how* generic.
16:55 Much of the theme information is actually stored in a nice generic way where modifying the files that exist is the right way to go.
16:55 "It's complicated."
16:55 Let me get back to PayPalling, but what I can promise you is that I'll check before assigning you things to make sure they're good for you to do.
16:56 ajonas k, sounds good.
16:59 paulproteus ajonas: I'm going to clean up these paypal-related patches and deploy them to the dev servers. Note that they operate in the PayPal sandbox, not the real PayPal world of cash flowing back and forth.
16:59 Then I'll explain where things are with that, and take feedback (-:
16:59 ajonas paulproteus: that is good
16:59 wouldn't want to have to pay to test!
17:06 xfxf left #miro-hackers
17:21 willkg afk for food....
17:21 paulproteus ajonas: If the site is payment overdue (which should only happen if the admin cancelled the paypal transaction, therefore it didn't auto-bill), okay if I redirect the front page to the login screen?
17:22 ajonas paulproteus: the front page for mc?
17:22 paulproteus For the MC site, yeah.
17:22 ajonas i don't understand, can you say more?
17:23 paulproteus If http://openhatch.mirocommunity.org/ has upgraded to an expensive tier, and starts a free trial, but then goes into PayPal and cancels the recurring payment before the trial is up, what happens on day 31?
17:23 ajonas they get downgraded to basic
17:23 paulproteus I say, http://openhatch.mirocommunity.org/ redirects to http://openhatch.mirocommunity.org/admin/ (and so we ask them to log in)
17:23 ajonas that doesn't make sense to me, it would be super confusing for the users
17:23 paulproteus The non-admin users, that is?
17:23 Okay, sure.
17:23 ajonas paulproteus: yeah
17:24 paulproteus Then I need to handle the case of site creation a little differently; I'll work on that.
17:24 ajonas paulproteus: but isn't this long after site creation?
17:24 paulproteus Yeah, but depending on the implementation, the two situations are equivalent for some parts of the code.
17:25 ajonas: One weird thing about site creation is that the way things are right now, you must log in to the site admin panel before you can do the paypal dance.
17:25 I guess you'd prefer things if the user could do the paypal dance without ever entering their admin credentials.
17:26 ajonas paulproteus: i am so confused, i thought that was the whole way we had set it up
17:26 so that you create a site, then from there go straight to paypal
17:26 then land back on your site
17:26 paulproteus Yeah -- I took a shortcut during implentation. I'll now undo the shortcut; sorry to bring it up again as if it's a new agreement.
17:27 ajonas i hate to make it more work, but i think the alternative is confusing.
17:27 paulproteus No prob
17:28 z3p paulproteus: I'm confused about http://git.participatorycultur[…]fa7987da621f8420a
17:29 paulproteus: why didn't context['request'] work?
17:29 paulproteus z3p: That's quite messy history. request.simple_context(takes_context=True) is only valid in Django 1.3.
17:29 I added it, thinking it would work, and pushed; then I discovered that was invalid in Django 1.2.
17:29 I should have kept that commit local and never let it see the light of day.
17:30 register.simple_tag(takes_context=True) # that's the thing that doesn't work in Django 1.2
17:30 I typo'd it in my IRC message in the past minute, confusing the issue further.
17:30 ajonas paulproteus: the disable custom domain email makes me smile.
17:31 paulproteus (-:
17:31 z3p paulproteus: I figured out what you meant
17:32 ajonas paulproteus: fyi: i'm going through all the emails now - i made a lot of changes to the wiki so the links need to be updated, and i'm going to change around some text.
17:32 z3p paulproteus: can't you work around that by making your own Node class which DTRT?
17:32 paulproteus z3p: Yes, except god I hate Node classes.
17:32 But: yes.
17:32 z3p paulproteus: sure, everyone hates Node classes
17:33 paulproteus: just trying to think of ways to get it working if Django 1.3 doesn't work out
17:42 kcw deardiary: rearranging bugzilla categories a bit; several categories don't make sense, and I've add Standard View, Metadata, Widget State
17:42 and in the process finding things that I should assign to myself
17:49 paulproteus ajonas: That's excellent.
17:49 afk a bit for eats.
17:56 janetPCF left #miro-hackers
17:57 janetPCF joined #miro-hackers
18:12 zanoi am i the only one here who doesn't like first run wizards at all?
18:26 willkg zanoi: i don't use them, but they're really helpful for most users.
18:27 zanoi for some reason i think that having stuff "just work" is much more helpful and attractive for users..
18:27 but i'm no usability expert so i could be wrong
18:28 willkg you're also probably not like most people.
18:29 zanoi heh, what's that supposed to mean :-)
18:29 it seems to me most linux programs are going away from first run wizards towards the "just work" direction
18:29 but i could be wrong there as well
18:30 willkg it's not a bad thing.  you have a lot more experience with applications and the jargon and ui things.
18:30 most people in the world don't.
18:30 zanoi for example music players will play what is in your ~/Music folders, web browser will per default just save into your Downloads folder etc.
18:31 i see it the other way around. Asking people where their download should be is jargon that confuses many new users
18:31 paulproteus I am in your camp, zanoi, fwiw.
18:31 willkg i don't think miro does that.
18:31 zanoi paulproteus: yay
18:31 willkg the first time startup pages are pretty few and just cover a few basic things to make it easier to use miro from the get go.
18:32 paulproteus It's possible that we're all in your camp. I don't remember the first-time questions anymore.
18:32 * paulproteus goes back to drinking his smoothie.
18:32 willkg the first one is "what language do you want the ui to be in".  that was a highly requested feature.
18:33 zanoi does firefox ask that on installation?
18:33 sorry, first run i mean
18:33 * willkg shrugs.
18:34 willkg the second one asks whether you want miro to start when you start your computer.
18:34 and the third is for finding media on your computer.
18:34 and that's it....
18:34 zanoi i see how changing language is a good feature, i don't see why it has to stop me from using the app for the first time instead of taking the default language
18:35 i'm not saying any of the features are bad, i'm just saying it might be confusing to users who just want to try out the app
18:35 * willkg shrugs.
18:35 willkg it was a highly requested feature.  i implemented it because of that.
18:35 zanoi sounds reasonable :)
18:36 willkg so...  i'm sure you can find people who don't like it and people who do.  frankly, i think it's a good thing to have because it fixed a bunch of problems with system ui language that people were having.
18:38 the defaults for the first time dialogs are such that you can click "next" "next" "next" "next" and you're done in less than 2 seconds.
18:38 i think we've done a good job of making first time startup user friendly without being annoying.
18:38 zanoi ok
18:39 willkg the cli frontend doesn't ask any questions.  maybe that's more to your liking.  ;)
18:40 zanoi heh.
18:40 from now on i will only use the cli then
18:40 sry, gtg afk dinner
19:00 DGMurdockIII joined #miro-hackers
19:06 bendk deardiary: working through my long email list
19:11 zanoi back
19:32 janetPCF left #miro-hackers
19:48 DGMurdockIII left #miro-hackers
20:05 bendk can you all stop working so hard please!
20:05 every time I go to check in a changeset, my repo is out of date
20:06 willkg heh.
20:09 bendk willkg: Just posted a comment to #15818
20:09 I want to also tell ponyo to try moving their database and starting fresh, but I'm not sure if that's the best advice
20:09 willkg bendk: 15818 is weird.
20:09 bendk what do you think
20:09 yeah, definitely
20:10 willkg i have no clue.  mostly i cc:d you because it seems like actual evidence that the data in the database was bogus.  i think prior to this it was a theory with no smoking gun evidence.
20:10 that bug makes me want to stick my fingers in my ears and go "la la la la la la la la".
20:11 bendk yup
20:12 willkg yay!  we're below the 60-P1-bugs mark again.
20:19 paulproteus willkg: Yay!
20:30 willkg almost have fast resume stuff done.  then i need to do some tweaking and then test it on OSX and Windows.  but it seems to be working nicely so far.
20:31 paulproteus willkg++
20:32 deardiary: Deployed an early but usable version of PayPal integration for Miro Community. Working on some other small issues, then winding down.
20:36 ajonas paulproteus: for paypal, it says "I authorize the merchant to use the available funding sources in my PayPal account based on PayPal's default preferences for the recurring payments above. PayPal will not share my financial information with this merchant."
20:36 paulproteus: can we give them the option of which account to use?
20:36 paulproteus Huh, probably. Can you file an individual ticket for that?
20:37 ajonas if it were me, i would not like that, because i might want to choose my credit card over my bank account or something
20:37 ok
20:40 paulproteus ajonas: I'm getting pretty exhausted, so I want to run the status of things by you, and then I should probably check out for a bit.
20:40 ajonas ok
20:41 paulproteus I think that the P1s and P2s can be cleared off by the end of Friday without much trouble.
20:41 ajonas paulproteus: fun!
20:41 paulproteus There's one big P1 that isn't filed yet, which is proper handling of *stopping* an existing recurring payment if you are transitioning.
20:42 ajonas i don't understand?
20:42 paulproteus Right now, we'd naively create a new recurring payment if you switch tier.
20:42 Hilarious, right? We need to also end the old one.
20:42 ajonas why is that bad?
20:43 paulproteus Because the old recurring payment would keep running for eternity also.
20:43 ajonas oh, i see
20:43 paulproteus So you'd be billed for all the tiers you've ever signed up for.
20:43 (-:
20:43 ajonas no good. we can't just switch the amount?
20:43 paulproteus We might be able to decrease it. Either way, I simply don't handle that case, and we should file a bug so I do.
20:44 ajonas ok.
20:44 paulproteus And I need to glue the pieces in place for testing site creation. That's probably a solid 2-3h of work. Then (I'm already gritting my teeth) I'm going to have to go through all the bugs for 1.2 and all the deployments and see which ones require configuration/dependency changes.
20:44 Dawnrell joined #miro-hackers
20:45 ajonas paulproteus: ok
20:45 paulproteus a couple other things - if you could take a quick glance at the email i sent from sam kaplan, that would be great. also, we still need to deal with the open ticket in zendesk
20:45 paulproteus That probably takes us to the end of Tuesday. And I'll look at Zendesk right now.
20:45 Let me brb first.
20:46 tty in ca. 20 min actually
20:46 ajonas k
20:52 willkg afk...  strapping baby to chest (which sort of makes me his mechanized suit of armor, i suppose).
20:54 kcw these database 'corruption' issues - I don't think they exist
20:56 do we do any cherrypicking-back of bugfixes for 3.5.1?
20:58 I'm going to tell Ponyo to hide the offending file from Miro
20:59 Dawnrell left #miro-hackers
21:14 bendk left #miro-hackers
21:20 willkg kcw: so that 15818 bug is really a bug in python?  if that's the case, then we should probably upgrade the windows build environment to python 2.6.6.
21:21 kcw willkg: I've push a workaround - as long as we only use the fileutils of get*time it won't be a problem
21:22 willkg kcw: the fileutils functions were all added for u3 support.  that's all pretty old and it's very likely the case that not all the code goes through fileutils anymore.
21:23 kcw: i'll toss in a bug to upgrade to python 2.6.6.  pretty sure it won't be too hard.  it's more annoying than anything else.
21:26 paulproteus Python 2.6.5 had an annoying bug that bit OpenHatch.
21:26 Oh, right -- it broke parts of Django, actually.
21:26 willkg mmm...  maybe i should look at the 2.6.6 release notes.
21:27 paulproteus http://code.djangoproject.com/ticket/13194 was the Django bug.
21:27 "Hilarious."
21:27 ajonas paulproteus: can i edit {% if tier_obj.permits_custom_domain %} to work for custom theme?
21:28 paulproteus ajonas: Yeah -- there's a similar version of that. You can find the list of all valid things after the dot by looking in localtv/tiers.py
21:28 kcw willkg: I just did - it's fixed
21:28 paulproteus There's a "class Tier" with a few methods, each of which start with "def ..." like "def permits_custom_domain()"
21:28 Odysimus left #miro-hackers
21:28 paulproteus I don't quite remember what the custom theme one is called off the top of my head, but it's probably "permits_custom_theme".
21:28 ajonas paulproteus: perfect, ty
21:29 bendk joined #miro-hackers
21:36 willkg kcw: i'm puzzled.  the bug you linked to talks about Python 2.5--not Python 2.6.  it was fixed way before Python 2.6 came out.  are you sure that's the right bug?
21:37 kcw willkg: that bug mentions in one of the last comments that it's still broken on windows; there's a note in the python 2.6.6 change log that says fixed problem with negative floats in datetime.datetime.fromtimestamp and added unittest
21:38 willkg kcw: really?  i don't see that comment anywhere.  the last thing on that bug is from 2007.
21:39 Odysimus joined #miro-hackers
21:40 willkg kcw: and the note in the python 2.6.6 NEWS file is from Python 2.6 alpha 1--not python 2.6.6.
21:41 kcw willkg: yeah, I was wrong about that... I'm trying out my test case in 2.6.6 on windows now
21:41 willkg kcw: ok.
21:46 kcw willkg: it doesn't work in 2.6.6 either. Looking back on the bug report, they said they were skipping the unittest on windows, where it still didn't work. I think windows can't handle negative timestamps in some functions, but can produce them in others
21:46 willkg ok.
21:47 i won't bother upgrading the windows build environment then.  whew.
21:47 kcw: does it make sense to add some code to the data integrity checking stuff in the database code to find and avert these issues?
21:51 kcw willkg: by the time they get to the database, they're datetimes - and it's converting to datetimes that we have problems. I think all we can do is use my fileutil workaround and be careful not to call os.path.get*time
21:53 morganknutson joined #miro-hackers
21:53 willkg kcw: ok.
21:54 kcw willkg: I find it a little disturbing that getmtime is returning a negative value on an OS where all timestamps are positive, but that definitely seems to be what's happening
22:00 It looks like it's just a matter of: getmtime returns the OS's actual mtime, datetime.datetime was designed for FAT32 when all dates were post 1980, datetime.datetime has not been updated to support the date range that is possible with NTFS. So really, the problem is in datetime, and we should be careful using it.
22:04 paulproteus kcw: Well, I'll be.
22:04 That's amazing.
22:05 kcw willkg: actually, we do store timestamps. We probably should put a check in database that sets any stored timestamp to max(timestamp, 0)
22:05 we also get timestamps from DAAP
22:06 and convert them to datetimes
22:06 willkg kcw: talk to bendk.  i think that would go in the databasesanity module, but i get fuzzy about how all that works.
22:07 bendk I think that code can be in SchemaDateTime.validate()
22:08 which gets called before we save things to the database (double-check that though)
22:08 if the value is out of range, we can change it and log a warning
22:08 hmm, actually wait a sec, I'm not sure if you can change things in validate
22:08 kcw we use fromtimestamp several times on values that haven't been in the db at all
22:09 bendk I think you would need to add that functionality as well
22:09 kcw maybe we should wrap fromtimestamp?
22:10 as far as I can tell it's only datetime that's broken, and that seems easier than finding all sources of timestamps
22:12 bendk sounds reasonable to me, I only have a few more minutes of time today and need to get in a fix quickly so I'm not following this as well as I should
22:16 willkg: okay, crash report dialog is almost perfect
22:16 zanoi deardiary: finally pushed the per display shuffle/repeat and saving, but now found a tiny bug....
22:16 bendk I did a fix on OS X to resize dialogs so that we didn't need your hack that reserves space for the hidden widget
22:16 also, now the scrollbars appear, but there's still a slight bit of funkiness
22:17 I'll fix that tomorrow, see you all later
22:17 zanoi bye bendk
22:17 bendk deardiary: went through email, fixed a couple bugs
22:17 bendk left #miro-hackers
22:30 arpu joined #miro-hackers
22:33 willkg left #miro-hackers
22:45 willkg joined #miro-hackers
23:05 zanoi deardiary: bug #9440 finally fixed!
23:10 maggie_s joined #miro-hackers
23:11 zanoi oh no, suddenly there are lots of bugs
23:12 willkg afk a bit....
23:19 ajonas left #miro-hackers
23:25 morganknutson left #miro-hackers
23:34 morganknutson joined #miro-hackers

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