[08:32] morning all! [09:37] dobey: correct - "the photos stuff on android doesn't sync. it only auto-uploads, as i understand" [09:42] karni: upvote bug #989506 [09:42] Launchpad bug 989506 in Ubuntu One Servers "API directory listing should filter out files with size=None (MakeFile but no Upload)" [High,Confirmed] https://launchpad.net/bugs/989506 [09:43] rye: I'm reading up on u1-internal, what you talked about with James. Is that it? [09:44] rye: ok, how do I upvote? [09:44] affects me? [09:44] karni: yep [11:21] finally back [11:21] rye, thx! [11:21] good morning! [11:24] gatox, hello! [11:24] mandel, hi [11:53] mandel, i forgot to ask you..... did you answer mmcc e-mail or i should do it? [11:54] gatox, I did no, I can do it, no worries [11:54] mandel, if you want..... i can do it too.... let me know === yofel_ is now known as yofel [11:55] hey, holas [11:55] alecu, buenas [11:56] alecu, bring me a t-shirt from UDS please!! :P [11:56] gatox, I'll try [11:57] mandel, last night I noticed that the last 3.0.0 installer still had qt 4.8 instead of 4.7 [11:57] mandel, so brian will make yet another installer today. [11:58] alecu, I read that, brian metioned py2exe being too smart and using the old build [11:58] it also had an older version of ssl than the one in 2.0.3 (0.9.8.12 in 3.0.0, instead of 0.9.8.18 in 2.0.3) [11:58] alecu, I have tested what we talked about, I done get ssl errors, instead the deferred is never called [11:59] mandel, .... what did we talk about? I remember we talked about many things, but not about that. [11:59] my mind is like a fortress [11:59] alecu, testing performing a request over https when we do not have ssl certs on the other side [12:00] mandel, and which deferred gets never called? [12:01] alecu, yield wc.request('https://blah.com') [12:04] mandel, oh, right. Well, I'm doing the same test, with a twisted server too, but with a qt client. [12:04] mandel, and I get different kinds of error, but nothing like the one we get on those logs. [12:05] hmmm... perhaps I should be running this tests on windows. [12:06] alecu, might be a thing to test.. [12:06] alecu, a platform might behave in a diff wa [12:06] y === salgado_ is now known as salgado === wgrant_ is now known as wgrant [12:25] lunch time here [12:25] * mandel lunch [12:34] dobey, if you have time for a C review: https://code.launchpad.net/~thisfred/u1db/number-mapping/+merge/103862 [12:55] good morning! [12:56] ralsina, hi [12:57] gatox, ping [12:57] alecu, pong [12:57] gatox: can you try this branch on windows? https://code.launchpad.net/~alecu/+junk/qt-ssl-error/ [12:58] ralsina, morning [12:58] alecu, ok [12:58] alecu, should i run the tests or something else? [12:58] ralsina, two bits of news: the latest version we gave joshuahoover mistakenly got built with qt 4.8 instead of 4.7 [12:58] oops [12:58] gatox: run the tests is fine. [12:58] alecu: qt 4.8 or pyqt 4.8? [12:58] ralsina, so brian will be making a new installer today with 4.7 [12:59] ralsina, qt [12:59] oooook [12:59] ralsina, also I found that the latest installer used a slightly older ssl version than the one in 2.0.3 [13:00] alecu: 2.0.3 was built in my notebook instead of brians [13:00] alecu: we could add checks for openssl version, I suppose [13:00] ralsina, (SSLEAY.dll version 0.9.8.18 in 2.0.3 but 0.9.8.12 in 3.0.0) [13:01] alecu, i get this: http://paste.ubuntu.com/949718/ do i need to add something to the path or anything? [13:01] ralsina, finally, I think I managed to reproduce the issue on windows by making a qt client connect to a twisted server with some broken certificates [13:01] alecu: "broken" how specifically? [13:02] gatox: that's awesome. [13:03] gatox: I'm getting this, which is much closer to the bug we are getting: [13:03] http://pastebin.ubuntu.com/949723/ [13:04] ralsina, so, in gatox's install, we are getting the rigth error, because the certificate is self signed. [13:04] ralsina, but in my install we are getting the "No error" and the certificate shows up empty. [13:04] How the HECK can error 0 be "the certificate is self signed and untrusted"???? [13:06] ralsina, in gatox install it comes correctly: the sslErrors gets two parameters: reply and a list of errors. The reply has error 0, but one of the errors (the only one) has error 9: The certificate is self-signed, and untrusted. [13:06] ralsina, that means that if gatox builds the installer, when it gets run on the clients machine it will log the exact ssl error. [13:06] on the other hand [13:07] in my install, the list of errors has one item too, but that one error is ZERO. [13:07] and that's completely wrong. [13:07] it should be error 9. [13:07] I suspect I have the same versions as brian. [13:08] ralsina, gatox: does this make sense? ^ [13:08] alecu: maybe [13:08] so, gatox: let's compare installed versions of qt and everything else. [13:09] gatox: go to Control Panel\Programs\Programs and Features [13:09] alecu: if it's the SSL, then we need to find a way to check the version and add it to the buildout or setup.py as a test and fail if we don't meet it [13:09] alecu, pyqt 4.8.4-1......... Qt 1.1.0 [13:09] qt 1.1.0 ? [13:10] alecu, mmmmmm that is wrong [13:10] wait [13:10] gatox: I've got PyQt v4.9-1 and pycrypto 2.3.0. [13:11] alecu, pycrypto 2.3.0 also [13:12] alecu, mandel, can you please review and run https://code.launchpad.net/~verterok/ubuntuone-client/fix-987376/+merge/103273 in windows? some stuff related to windows was modified there, thanks! [13:12] alecu, Qt 4.7.1 [13:12] alecu, i didn't update my windows vm it seems [13:13] gatox: DON'T [13:14] gatox: that's great :-) [13:14] gatox: finding Qt 4.7 + PyQt is not easy ;-) [13:14] mandel: was sso the only thing that used txtcpserver? [13:14] I'm checking in an older vm, too. [13:15] It's got PyQt 4.8 (probably qt 4.7) [13:15] lo and behold, I get the same error as gatox. [13:16] facundobatista: nothing to worry about, not it's the same in windows and linux :) [13:16] verterok, I was trying to get a free review! [13:17] hehe [13:17] facundobatista: yes, it should be tested in windows too [13:18] alecu, mandel, there, there, :) [13:19] facundobatista, verterok: you are very lucky that we have review days, and that gatox and/or dobey will be able to run the tests on windows! [13:20] well, probably not dobey :P [13:20] gatox: ^^ :) [13:20] ok..... running the tests..... [13:32] verterok, facundobatista http://paste.ubuntu.com/949773/ [13:35] gatox: weird, the change is unrelated :( [13:35] verterok, i'm running trunk now, and i get the same error [13:35] ouch [13:35] gatox: is jenkins not reporting the same error? [13:36] dobey, i don't know [13:36] hrmm [13:40] gatox, you broke it [13:41] facundobatista, ?? [13:41] hrmm, jenkins seems to report a few other errors [13:43] facundobatista, mmmmm it seems related to a branch i propose...... but it was working in my branch...... [13:45] alecu: so i should expect a new installer today to send out? [13:46] dobey, yes :) [13:46] facundobatista, looking :) [13:47] gatox, connection test case error? wtf? any idea? [13:47] joshuahoover, yes: as soon as briancurtin wakes up we'll ask him to work on that. [13:47] alecu: heh, you haven't called him yet? ;) [13:48] alecu, joshuahoover, shall I send the flying monkeys to chicago? [13:48] mandel: yes please [13:48] mandel, nop..... i'll look for the branch that was related to that to see the change (i'm finishing with the bloody mac scripts for tests rght now)....... but i know it was working when i proposed it [13:48] alecu: i'm guessing you're talking about the versions? i'm looking right now [13:48] mandel: is that where they keep the flying monkeys - spain? [13:48] it's weird..... [13:49] mandel, those are some fast monkeys! [13:49] joshuahoover, yes, we also have rats :) [13:49] briancurtin, awesome! [13:49] gatox, hm, let me try and run them on my vm [13:50] briancurtin, so, for the record: this morning I made a small tests branch that reproduces the ssl error that we found on the logs. [13:50] alecu, can I see that one? [13:50] briancurtin, I managed to make it fail using pyqt 4.9 (qt 4.8) on windows. [13:51] briancurtin, but it worked for gatox, who has pyqt 4.8 (qt 4.7) [13:51] mandel, https://code.launchpad.net/~alecu/+junk/qt-ssl-error/ [13:51] interesting [13:52] mandel, take a look at gatox's run: http://paste.ubuntu.com/949718/ [13:52] mandel, and compare it to mine: http://pastebin.ubuntu.com/949723/ [13:52] (btw, just noticed my test run took 30s longer than gatox's) !!!! [13:53] alecu, hm, is yours the one with no cert details? [13:53] mandel, right! [13:53] mandel, and No error (0) [13:55] mandel, so, we get the sslErrors(reply, errors) signal. reply is still going on, so it has no error "Unknown error (0)". [13:55] alecu, so, with Qt 4.8 you get no error (0) and no cert details, gatox gets (0) but also (9) which is for a selfsigned cert [13:55] mandel, but the "errors" list has one item, with "No error (0)" (WRONG!) and empty certificate (WRONG!!!!) [13:55] mandel, right. [13:56] mandel, gatox is using qt 4.7 [13:56] If we isolated it this far, we can probably do a good bug report for upstream [13:57] alecu,wow! so, ignoring the fact that we are getting a (0) with gatox which I cannot explaing, 4.8 has a serious bug qhen dealing with ssl [13:57] i feel that everything works different for me always :P [13:57] mandel, I can explain the first (0): it's the expected [13:57] behaviour. [13:57] alecu, why? [13:58] mandel, at the point the sslErrors signal is thrown, the "reply" still has not decided if it will succeed or not. [13:58] mandel, it's up to your handler code in sslErrors to decide if you want to ignore the certificates or not. [13:59] mandel, if your code decides to ignore the certificates it can do so with .ignoresslerrors or something like that. [13:59] alecu, and due to that it will always have a first (0) in the list? [13:59] mandel, no. In gatox's case the first (0) comes from "reply", not from "errors". [13:59] alecu, oh, ok [13:59] mandel, remember: signal sslErrors(reply, errors) [14:00] mandel, and only "errors" is a list. [14:00] alecu, I though I was looking just at the list of sslErrors [14:01] mandel, can you take a look at the code in that branch and try to run it? [14:01] alecu, yes, I'm reading it right now [14:02] ralsina, if you have pyqt 4.9 in your vm, please try running that branch too. [14:02] ralsina, https://code.launchpad.net/~alecu/+junk/qt-ssl-error/ [14:02] alecu: no windows VM, would be a reboot [14:02] and briancurtin too. [14:02] alecu: and I'd rather not today :-( [14:02] ralsina, no prob. [14:03] alecu, what happens if you use the DefaultOpenSSLContextFactory? [14:03] mandel, dobey could you review this branches? https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/mac-port/+merge/103488 - https://code.launchpad.net/~diegosarmentero/ubuntuone-windows-installer/mac-env/+merge/103525 (If you are going to test it IRL, remember that you need to run first the env-mac script from the u1-windows-installer branch) [14:04] does it really make sense to have mac branches of ubuntuone-windows-installer? [14:05] duanedes1gn: ping [14:06] dobey, no, it does not, I would do a new project for that [14:06] mandel, it's the same if I use the Default.... [14:06] alecu, so, why the diff? [14:06] mandel, I just subclassed it so I could play with including/excluding some bits of the certificate. [14:07] alecu, also, here we found a bug in the fact that the cert info is not given, yet not why it happens with our 'correct' certs, right? [14:07] mandel, "it's the same" -> "it should be the same". Because I've not tried it. :-) [14:07] alecu, hehe, I'll try it :) [14:08] mandel, dobey with those branches, it's possible to execute the tests in mac without ttouching anything in u1-client, or putting unnecessary things in the environment of the user [14:12] alecu: ok, wow, with the proper older PyQt/Qt setup, u1cp does not start at all. nothing in log files about it either :/ [14:13] alecu, I see the same behaviour as you do, that is, no cert details on my windows vm, on P I can see the cert details [14:13] gatox: why do you dump stuff into ~/.profile instead of just altering the current environment? [14:14] alecu, so, an other options is pushing the version to be the same as the on in P [14:14] dobey, like what [14:14] ? [14:14] alecu, but as briancurtin is a PITA [14:14] dobey, what do you mean with "altering the current environment"? [14:15] mandel: if we really need that im sure i can figure it out eventually, but i spent almost a full day and got nowhere with it, and it seems no one tries to build their own PyQt with a custom Qt version [14:18] dobey, you only execute the env-mac script once..... so i need to save the references to u1lint, etc in some other place to be accesed for the scripts in each project [14:19] briancurtin, alecu, so, if we go back on version, it does not start yet we see the cert info, if we move up the qt version in theory we get both.. [14:21] this PyQt/Qt I have installed doesn't have QSslCertificate, apparently ("from PyQt4 import QtNetwork" works, but it doesn't have QtSslCertificate) [14:21] gatox: i mean, why doesn't it modify the environment when run, and why don't you just have it executed from ~/.profile? [14:22] briancurtin, wow! [14:22] hi all, I bought an album in the music store (rhythmbox 12.04) yesterday, and it hasn't been downloaded yet (either to my machine, nor to the web streaming player) is there a way to diagnose the issue further? [14:22] gatox: rather than just blindly dumping extra stuff at the end of .profile [14:22] s/QtSsl/QSsl/ on that last line [14:23] dobey, i'm not sure i understand you..... i can't just modify the environment when i run the tests...... because the run-tests script from sso.... doesn't know about the buildout [14:23] joshuahoover, rye, duanedes1gn: ^^ can you look at james_w's album issue? [14:23] dobey, gatox, I really think we should create a mac project for this, adding it you windows is not a good idea [14:23] thanks dobey [14:23] gatox: no no, not in run-tests. in this env-mac script [14:23] gatox: i never mentioned run-tests [14:23] mandel, maybe we should rename the windows one to something generic [14:24] dobey, example? i'm not following you [14:24] gatox, AFAIK there are windows things there, right? [14:25] mandel, yes...... but doing another project for mac..... we are be copying a bunch of stuff from there....... so we are going to have duplicated things..... and we will need to mantain two simmilar projects [14:25] gatox: have your .profile run the env-mac script, and have the env-mac script do the export FOO=$BAR stuff, and it gets executed every time you log in, and you don't have to re-log (or restart the terminal) to get a working environment. [14:25] mandel, gatox: we can deal with the project naming issue later [14:26] dobey, i'm not sure i like that solution [14:27] eventually, and i dont know when this will be, but this whole buildot thing will probably live in its own top level project, not in a subfolder of ubuntuone-windows-installer [14:27] gatox: and dumping junk in your ~/.profile is a better solution? [14:28] briancurtin, mandel: sorry, I was on mumble. [14:28] catching up with the log. [14:29] dobey, i'm not sure about running a script that finds stuff and that, everytime you open a terminal [14:29] hi everyone, catching up === mmcc_absent is now known as mmcc [14:30] gatox, what dobey is saying is the same as what I ended up doing on Tuesday [14:30] czajkowski: pong [14:30] except I didn't put env-mac in my profile because I'll forget it's there [14:31] gatox: why are you doing a find? what happens if the path changes? then your ~/.profile is broken, or you run the script again and you have a bunch of duplicate junk at the bottom of it [14:32] gatox: or why not just run it whenever you need it, and it only modifies current environment, instead of injecting stuff into .profile [14:32] dobey, gatox, I much more prefer the last option ^ [14:32] dobey +1 that;s what I like too, for the reasons I said above [14:32] dobey, gatox, I really don't fill good on my env being modified to run tests and be left like that [14:34] not modifying .profile at all??...... so you need to searchhhhh for the env-mac script everytime you want to do anything u1 related? [14:34] i'll prefer to add at least an alias in the .profile to run the script [14:34] that's how i do it on windows with that env.bat script. if you want to do u1, start up a u1 env [14:35] gatox: well you can add an alias to your own .profile, and others can just run it when they need it [14:35] gatox, you can always put an alias, but just don't do it automatically [14:35] * mmcc will let dobey type for him for a while [14:36] ralsina, 1-1 before standup? [14:36] mandel: sure, starting mumble [14:36] heh [14:40] briancurtin, so, what version of PyQt did you install? I find it really weird that it has no SSL support. [14:40] briancurtin, perhaps it needs pycrypto installed too? [14:41] if it did, it would be nice if it told me that. i *just* uninstalled it 1' ago, i'll reinstall to get the proper version numbers [14:41] i couldn't run your test because the QSslCertificate import failed, but QNetwork (which ssl is in) works fine [14:42] awful. [14:42] mmcc, can you share your script? [14:43] gatox, sure, which script are you referring to? The mac-tests script I did for u1-client, or ? [14:43] mmcc, no, the one which you use yo setup the environment [14:45] ok, just a sev [14:48] gatox: http://paste.ubuntu.com/949918/ -- it is just the output of your script, run for me, with some tweaks [14:49] alecu: i'm on PyQt 4.8.6 - Qt 4.7.4. QSslCertificate was added to Qt in 4.3. there's no QSslCertificate in the QtNetwork package [14:49] gatox: if I was going to put it in the buildout I would just use `pwd` and relative paths, i guess [14:49] briancurtin, can you try installing pycrypto? [14:50] mmcc, no, i didn't want to add that..... just wanted to see how are you doing the script...... because if i execute a script that contain that...... as soon as the script finish, everything that the script set, is not remembered (which is correct for me) [14:52] mmcc, so, how are you executing that? [14:52] gatox, oh, I just do "source scriptname.sh" [14:52] applies changes to current environment [14:54] mmcc, thanks [14:54] gatox, sure. I think in bash you can use '.' as a shortcut for source. that works in zsh too. you know, if 'source' is too many letters ;) [14:58] alecu: no difference there [14:58] mmcc, briancurtin, gatox, mandel, alecu, dobey, thisfred: standup in 2', prepare notes [15:00] me [15:00] me [15:00] me [15:00] me [15:00] me [15:01] what's with all this preparing? Ludicrous speed… go! [15:01] me [15:01] thisfred: you? [15:01] me [15:02] mandel, go, thisfred is last [15:02] me [15:02] DONE: fixes to support domain sockets. Continued work on tcpactivation, some ssl debugging with alecu. A bit of FF. Reviews to gatox branches. [15:02] TODO: finish tcp activation, fix jenkins ec2-windows node, re-review gatox branches, administration stuff (swap days etc..) [15:02] BLOCKED: no [15:02] gatox, please [15:02] DONE: [15:02] Mumble with mandel and mmcc yesterday, finishing with the last modifications of the mac scripts. Started with the platform refactoring, mumble with lissette about a styling issue. [15:02] TODO: [15:02] Propose the last version of the scripts, keep working in the refactoring, fix some tests in u1-client [15:02] BLOCKED: [15:02] No [15:02] ralsina, go [15:02] DONE: reviews, team call, 1-1s, other calls, helped around. TODO: implement a website, because it's friday and I am really trying to get rotated into beuno's team (and also other reasons) BLOCKED: django is a little more annoying than I remembered, but no NEXT: briancurtin [15:02] DONE: got individual pieces of the installer functionality working...but bitrock doesnt make it very easy to tie them together [15:02] TOOD: figure out how to tie this installer together to fix the upgrade case and make it actually work properly [15:02] BLOCKED: none [15:02] NEXT: alecu [15:02] mandel: when you say "gatox, please" I picture the yao ming meme. Now you do too :-) [15:03] DONE: various mumbles, found version mismatches and managed to reproduce SSL error on windows [15:03] TODO: more SSL debugging [15:03] BLOCKED: no [15:03] NOTE: national holiday on monday and tuesday [15:03] NEXT: mmcc [15:03] done: got bug/branch/merge workflow working, installed precise, downloaded win7, looked around at mac alternatives for pyinotify, [15:03] todo: finish up bug fixes, install windows, run tests on all OSs [15:03] blocked: none [15:03] uh [15:03] DONE: bug #987412 bug #987414 TODO: bug #987412 BLOCKED: no [15:03] Launchpad bug 987412 in U1DB "support indexes on non-string types" [Wishlist,In progress] https://launchpad.net/bugs/987412 [15:03] Launchpad bug 987414 in U1DB "Implement get_index_keys_from_index" [Wishlist,In progress] https://launchpad.net/bugs/987414 [15:03] ok [15:03] λ DONE: backports, review [15:03] λ TODO: reviews, backports, SRUs, u1db packaging/buildsys [15:03] λ BLCK: none. [15:03] lol; [15:03] thisfred: early! [15:03] since alecu was temporarily blinded or something [15:04] or something. Sloppy :-) [15:04] dobey, sorry! :P [15:05] what are you guys talking about, seriously [15:06] mmcc: the standup was out of order [15:06] mmcc: we have a running problem with having all of us stand in line and type a status. [15:06] ralsina: oh ok this would be helpful: what is the order? :) [15:07] mmcc: whatever order we say "me" in [15:07] ralsina: gotcha [15:07] mmcc: in this case, alecu broke the chain by passing the baton to you instead o dobey [15:08] As for the yao ming mention: http://memegenerator.net/instance/19536695 [15:08] mmcc: also, this team is ... somewhat informal. [15:08] mmcc, you mentioned " mac alternatives for pyinotify [15:08] " [15:08] ralsina :) [15:08] haha [15:08] 'somewhat informal' [15:09] * ralsina is probably a bad influence [15:09] mmcc, we've been discussing some of the alternatives a few weeks ago. [15:09] hello! [15:09] why does u1 hate me :( [15:09] alecu: yes, I was up late, so I poked around. I figured, if pyinotify could be easily ported or replaced without much change in our code, that'd be good. [15:09] roasted: did you kick it when it was down? [15:09] mmcc, this is a nice introduction to the problem of filesystem events in osx: http://arstechnica.com/apple/reviews/2007/10/mac-os-x-10-5.ars/7 [15:09] ralsina, heheheh [15:09] roasted: it doesn't hate you, it may not understand you! What happened? [15:10] It comes up every so many minutes and says "Ubuntu One is now downloading 200 files"... minutes later, "Ubuntu One is now downloading 199 files" [15:10] I think it's confused from what I did to it yesterday :( [15:10] mmcc, it does not look likely that pyinotify could be easily ported. [15:10] roasted: that's a known bug [15:10] roasted: did you dump more than 200 files into it? [15:10] roasted: it maxes at 200 files. You probably have much more than 200. [15:10] I do have more than 200. I think 260ish. [15:10] I synced my android to U1 (backing up pictureS) and I hoped it would sync down to my computer. [15:11] roasted: the numbers should eventually start making sense [15:11] mmcc, so, all osx notifiers are based on the FSEvents API discussed in that article. [15:11] For an automagical download of my cell pictures [15:11] roasted: every time a file in that finishes syncing, it will notify of "X and 199 other files" until you have less than 200 left :-/ [15:11] mmcc, but FSEvents does not provide "per file" notifications. [15:11] roasted: it's a known bug [15:11] alecu: ok, didn't look at pyinotify code, but I found this: http://pypi.python.org/pypi/MacFSEvents/ [15:11] alecu which claims to provide per file notifs [15:11] alecu: and pyinotify style interface, down to the cookies [15:12] alecu, it starts separate threads to watch things though, so that's a difference. [15:13] mmcc, looking [15:13] dobey: so, what should I do? are you suggesting it'll *eventually* sort itself out and sync everything to my computer? [15:13] roasted: yes, no action required on your part [15:14] * briancurtin need coffee, back shortly [15:14] ralsina: ah okay. I just didn't think it would take a while. I left the igzmo run for a good 30 minutes and didn't see any new pictures in that folder in my home directory.... [15:14] roasted: yes, you'll just get some extraneous notifications [15:14] roasted: file sync may be a bit slower than normal currently, due to the Ubuntu 12.04 release [15:15] dobey: hm, I guess I didn't think they would be linked... the release + U1 that is. But that makes sense. [15:15] mmcc, this bit sounds expensive: "Note that the logic to implement file events is implemented in Python; a snapshot of the observed file system hierarchies is maintained and used to monitor file events." [15:15] dobey: I suppose U1 goes to the main server, unlike the repos where I can "select best server" for higher speed? [15:15] ha, imagine that. I JUST had a new picture pop up in my folder. [15:16] alecu, I saw that too. bit of a red flag, but I wouldn't know for sure without testing it [15:16] roasted: indeed we don't have 'mirrors' for u1 :P [15:16] gotcha :D [15:16] btw, what's the story on licenses? this thing is BSD licensed. can we use that? [15:17] (at my last job we were regularly reminded to be super paranoid about software licenses, as you might imagine) [15:17] mmcc: licenses are hard. if it's plain BSD without any insanity, it should be fine to use [15:18] dobey, from pypi it is listed as straight bsd, so I expect no unusual insanity [15:19] I am sure its been asked a million times but is there a timeline for getting an ubuntu1 app for mac? [15:20] it's being worked on [15:20] steveccc_: mmcc was hired last monday to work on it, along with several other people in this very channel :-) [15:21] i appreciate they are working hard but does that indicate days, weeks, months etc [15:21] steveccc_: months [15:21] ok many thanks [15:21] sometime before the big rip :) [15:22] i prefer to work on linux but work machine is mac and hence just trying to pick the best external storage for the moment - think will have to be dropbox for now [15:26] steveccc_: if you don't mind saying, what kind of work do you do that requires a mac when you prefer linux? [15:26] mmcc, I'm taking a look at the code from MacFSEvents. I like that it provides an api similar to pyinotify. Also, as I expected, it does a stat on every file in a given dir every time it gets a "something changed in this dir" event. [15:26] ok, need to get lunch. bbiab [15:27] mmcc, since that sounds like it would be painfully slow, we've been considering the option of having a separate small daemon running as root, and taking events from /dev/fsevents [15:28] mmcc, that is, if the user allows that bit to be installed as root. If not, we would revert to some of the ways to use FSEvents. [15:28] alecu, why does it have to run as root? [15:28] mmcc, because that's the way /dev/fsevents works :-( [15:28] mmcc, it's well explained in the ars article I pasted above. [15:29] alecu, ugh. but I guess if you use the fsevent C api, you don't need to be root? [15:29] I'm more than happy to write a C extension to avoid having to run as root [15:29] mmcc, if you use the FSEvents api you don't need to be root. But you don't get "per file" events. Only an event saying "something changed in this folder" [15:29] mmcc, and that's what MacFSEvents does. [15:30] mmcc, also the "watchdog" python module, and a few more. [15:30] alecu, ok, now I get what you were saying. [15:30] about stat() - it gets a dir event then stats the files to figure out what changed. [15:30] right? bad. [15:31] mmcc: unfortunately have to use adobe CS3 [15:31] mmcc, right. FSevent provides the path where changes have happened to the callback. And it's up to the callback to stat every file to see which file changed. [15:32] mmcc, that's why it needs to have the previous stat of each file. [15:32] mmcc, I see that MacFSEvents uses the file time. But we should also keep a checksum, to be extra careful. [15:33] steveccc_: ah, that makes sense. so I assume you have seen http://adobegripes.tumblr.com/ ? fi you get nothing else from our conversation, I hope to lend a chuckle... [15:34] alecu, yes it seems like a bad option after all [15:34] mmcc: thanks - will have a look later - have to disconnect now - thanks for the help [15:37] briancurtin: any estimate on when a new build with the older qt will be ready? i'd like to send that out asap in hopes of getting some feedback yet today [15:38] joshuahoover: unfortunately never, unless i find a way to build my own custom package of it. the older PyQt/Qt package that I downloaded does not contain SSL support for whatever reason... [15:39] but alecu might have a better handle on what's needed since he reproduced the issue on the current version [15:39] briancurtin: ah, ok :( [15:39] ralsina: soooo...i'm not sure what we should be trying next but we need to keep eliminating possibilities between why 2.0.3 works for these impacted users but not 3.0.0 [15:39] mmcc, so, we were planning on having a daemon listening for file events. Either one running as root (reading from /dev/fsevents), or, if the user did not allow it, we would have our syncdaemon start one that reads from the FSEvents API, and doing something similar with stat and checksums to what we discussed. [15:40] alecu, ok, I understand the plan. [15:40] mmcc, we were planning to use the "watchdog" module. But MacFSEvents looks nice too. [15:41]