[08:38] Good morning all! :) === zyga-afk is now known as zyga === yofel_ is now known as yofel [11:08] good morning! [11:16] gatox, morning! [11:17] gatox, how are things in the fsevents world? [11:17] mandel, i have to fix a couple of bugs, dirty reactors things and stuff like that (clearly not everything was refactored correctly), but after that is done [11:19] gatox, cool, I'm dealing with the encoding of darwin because we cannot use the unix methods.. I'm moving out all the decorators and renaming them to ospath etc.. later ospath will do the right thinig for windos and for darwin and that way we have a single code base for them [11:19] gatox, I might move the unix.py back to linux because it does not make sense to have linux.py and unix.py [11:20] mandel, ohh that's cool, right now we are just assigning lambda to that decorators on mac (in my branch) [11:23] gatox, this is for os_helper atm.. [11:23] gatox, and the file system notification that uses them [11:24] gatox, I'm moving from is_valid_windows_path to is_valid_os_path for example [11:24] whihc makes more sense in a way.. [11:24] mandel, yes..... i'm doing EXACTLY that on the refactor too [11:25] Hi. Is the web client considered part of ubuntu-client, or another project? [11:25] mandel, but not assigning any behavior for mac.... just return the same function [11:25] gatox, I should have a partial fix (just with the refactor in a few mins) so maybe I can propose and you can review it and try to merge it [11:26] Talliesin, AFAIK is ubuntuone-servers or something of the kind [11:26] mandel, yes!! that would be save ugly ifs to avoid pyflakes issues [11:26] gatox, working as fast as I can without making everything wrong :) [11:27] mandel, no worries..... i have stuff to fix [11:27] gatox, with the darwin encoding fix I'll try to sync again my machine, atm is stuck due to local rescan failing.. [11:27] what a set back to my plans ;) [11:27] :P [11:30] mandel: Thanks. Other bugs seem to suggest that. Best of luck with your darwin work. I'm looking forward to getting my mac-using partner onto it. [11:32] Talliesin, thx! we should get there in a few months (that means packaged etc..) [11:44] Sweet/ [12:21] hola everybody! [12:22] alecu, morning! [12:22] alecu, so, semi bad news from yesterday: stackoverflow.com/questions/9757843/unicode-encoding-for-filesystem-in-mac-os-x-not-correct-in-python [12:22] AlanBell, hi! [12:22] alecu, found it by running sd on my music collection and fix in it atm [12:23] classic gatox, says hi to someone else [12:23] :) [12:23] jaaaaaa [12:23] AlanBell, sorry..... [12:23] alecu, hi! [12:25] :) [12:25] well hi to you to anyway o/ [12:25] :P [12:26] mandel, is a technique to make new friends :P [12:26] jejej [12:27] gatox, I should try that at the bar.. [12:27] mandel, jejeje [12:28] mandel: I think this is new unicode issue is a tricky one. [12:28] mandel: what's your proposed solution? to always normalize after reading from the local filesystem? [12:29] remind me, do we store the paths in unicode? [12:29] or as bytes? === bac` is now known as bac [12:31] Chipaca: do we store them... where? [12:31] Chipaca: inside syncdaemon all paths are bytes. [12:31] alecu: everywhere, i hope :) [12:31] i mean, i hope we aren't converting to unicode on the server and bytes on the client [12:31] utf-8 bytes, that is. [12:32] right [12:32] osx normalizes unicode before converting it to utf8 [12:32] (as everybody should, but don't) [12:32] right [12:32] unfortunately, because they're them, they normalize with something that isn't a standard normalization [12:33] Chipaca: and we are (very likely) blindly comparing bytes, so we need special care with normalizations. [12:34] * alecu wonders if unicode.__cmp__ knows about normalization. [12:34] alecu, Chipaca, exactly, so the idea is to normalize it before we send it to sd, we need os_helper to deal with it like we do on windows [12:34] alecu: it does not [12:34] alecu, my idea is to make the decorators do the right thing and share the implementation [12:34] unicode support in python is quite incomplete [12:35] mandel: you are simplifying things! we don't know if the utf-8 inside syncdaemon are normalized or not. They can be both and it would fail. [12:36] the normalization is NFD, but only if the character is not in the ranges: 0x2000-0x2FFF, 0xF900-0xFAFF, 0x2F800-0x2FAFF [12:36] ok, I bloody hate file systems + unicode + python [12:36] (apparently; haven't tested this) [12:37] alecu, shall we mumble about this after my lunch? [12:37] mandel: sure [12:38] alecu, ok, then I'm off to have lunch [12:39] Chipaca: do you have a reference to that bit of fact? [12:39] alecu: the link i had is now broken [12:40] * Chipaca is searching [12:40] http://developer.apple.com/library/mac/#qa/qa1173/_index.html [12:40] alecu: ^] [12:43] Chipaca, alecu, is also in the bug I added to u1-client regarding this with the u1-darwin tag (I mean the qa url) [13:09] good morning! [13:09] hola ralsina! [13:15] ralsina, hi [13:15] hola gatox, alecu === jdobrien is now known as webm0nk3y === webm0nk3y is now known as jdobrien [13:50] alecu: we'll have to reschedule your 1-1 because it now collides with another call, can we move it to 11:15 [13:50] ralsina: sure [13:55] alecu, mumble whenever you are free :) [13:56] ralsina: I'm starting a mumble with mandel. [13:56] alecu: then let's do it after the weekly call [13:56] ralsina: great [14:00] alecu, can you hear me? [14:01] alecu, are you speaking? [14:01] mandel: I can't hear you at all. [14:01] alecu, por que se llaman los supositorios asi? [14:01] alecu, lets me restart the bloody app [14:01] mandel: I was speaking, all the time. [14:01] mandel: supongo que... no se! [14:02] alecu, por que supon que te gustan :P === zyga is now known as zyga-food [14:14] mandel: jojojo [14:14] alecu, jeje [14:14] mandel: let's switch to here, mumble is sucky for me today. [14:14] alecu, yes, please! [14:15] alecu, that is what I said over mumble, I don't know if you got it.. [14:16] @ping [14:17] alecu, might be my internet connection :( [14:17] oh, no hal here.. [14:17] mandel: @pong [14:18] mandel: I'm trying to see if my router knows about something that might be eating my upstream bw [14:18] mandel: can we try mumble a little bit more? [14:19] alecu, so, the summary is, lets do listdir with the normalization and do nothing for the rest of the methods, mention in the weekly meeting the issues of the unicode + normalization [14:19] alecu, sure, lets me launch it again [14:19] mandel: are we running into files we don't find because they are in another normalization on disk? Really? [14:19] mandel: now I see the lips turning red, but I can't hear a thing. [14:20] alecu, just said things [14:20] ralsina: osx automatically does some normalization. [14:20] ralsina, we get encoding errors in the local rescan [14:20] mandel: I'm declaring my mumble broken. [14:20] alecu, +1 [14:20] mandel: let me try the one in the laptop. [14:20] alecu: so does linux, they do different ones though [14:20] alecu, ack, I'll me in mumble [14:20] ralsina: wha wha what? [14:21] ralsina: where does linux the normalization? [14:21] alecu: let me find the references [14:21] ralsina: I always thought linux considered the filenames as "just bytes" === m_conley_away is now known as m_conley [14:21] dobey, ralsina, is this ok with us: https://github.com/stig/json-framework/blob/master/LICENSE.md [14:22] alecu: "the preferred way of encoding text in Unicode under Linux should be Normalization Form C" [14:22] alecu: from http://www.cl.cam.ac.uk/~mgk25/unicode.html [14:22] mandel: looks good to me [14:23] ralsina, oh.. I removed the lib to use one from apple but it is not supported in 10.6 and we want to support 10.6 right? [14:23] alecu: hrm wrong doc, I have not seen this in a long time [14:23] alecu, cannot hear you [14:23] mandel: yes we do [14:23] mandel: check your volume [14:23] mandel: and your plug :-) [14:24] ralsina, everything works in my machine.. in theory [14:24] ralsina, can you jump in mumble? to test it mainly [14:24] mandel: sure [14:24] alecu, mandel: here's Ned Batchelder on normalizaion of fnames, he knws best, usually ;-) http://nedbatchelder.com/blog/201106/filenames_with_accents.html [14:25] I hear you both [14:27] ahhhhh..... refactor from hell!! [14:27] * gatox feels better now [14:29] ralsina: 3.0.2 released and will be picked up by autoupdater...waiting on them to update the http://one.ubuntu.com/windows/release page to 3.0.2 (still shows 3.0.1) === zyga-food is now known as zyga [14:29] briancurtin: remind them just in case, they usually forget that one [14:29] yeah i sent it on the ticket as a last step before we're done [14:36] mandel, alecu: [14:36] -rw-rw-r-- 1 ralsina ralsina 0 Jun 21 11:36 lé.txt [14:36] -rw-rw-r-- 1 ralsina ralsina 0 Jun 21 11:35 lé.txt [14:55] yes! refactoring complete [14:57] dobey: see comment 5 for bug #997326 about the valicert on ubuntu for turkish users ... MassiveTR confirmed that fixed the problem for him...i'm not sure why [14:57] Launchpad bug 997326 in Ubuntu One Client "Users in Turkey can't connect due to invalid Valicert certificate" [High,Triaged] https://launchpad.net/bugs/997326 [15:01] joshuahoover: awesome. :( [15:01] looks like mmcc is having problems, might not be in mumble [15:01] dobey: uhhh...yeah [15:01] dobey, mumble [15:01] thisfred: mumble [15:18] ralsina, alecu, briancurtin: delete is not supported on os x https://bitbucket.org/kang/python-keyring-lib/src/59a11c6cb548/keyring/backends/osx_keychain.py and only supported on win 7 on windows https://bitbucket.org/kang/python-keyring-lib/src/59a11c6cb548/keyring/backend.py#cl-786 [15:18] * mandel hates pykeyring [15:27] ok, need to get lunch now. bbiab [15:31] mandel, alecu review please: https://code.launchpad.net/~diegosarmentero/ubuntuone-client/darwin2-fsevents/+merge/111427 [15:34] Hi everyone. Not done any development with Ubuntu One before and don't have much experience with it at all. First, am I right that you can store contacts on Ubuntu One? And secondly, if so, is it possible to grab these out via an API? [15:36] * gatox lunch! === gatox is now known as gatox_lunch [15:36] I am building an application for Ubuntu and would like to connect to Ubuntu One, pull down the contacts and then find if any of those email addresses are also using my application (so people can connect, a la Facebook, etc). [15:37] dotty: maybe beuno can explain that [15:37] Thanks ralsina [15:55] ralsina, alecu I'm eod a little earlier today to try and fix the bloody internet connection.. [15:55] this thing is driving me crazy.. [15:55] Hi, is there still a problem with U1 and Picasa upload? [15:55] mandel: it's the blood. It's not a good conducting material. [15:55] mandel: try copper wire instead. [15:56] ralsina, meh, que chiste mas malo! [15:56] scarleo: doesn't ring any bells, what problem are you experiencing? [15:56] mandel: they are bad, but they are mine! [15:56] ralsina, hahahaahha see, that one was better :) [15:57] ralsina, when u1 is installed gallery wont show any albums when trying to upload to picasa [15:57] so it fails [15:57] scarleo: what gallery? Sorry, I may be missing some context [15:57] ralsina, photo gallery on Android phone, it has an upload feature to upload to Picasa [15:58] scarleo: ok, got it now :-) [15:58] scarleo: karni may be able to help you [15:58] I discussed this with someone on here and did some testing but it was a few months ago [15:58] scarleo: remember the name? [15:58] ralsina, sorry, no [15:59] scarleo: it was probably karni or rye, let's hope one of them can give you a hand [15:59] * karni reads [15:59] ralsina, think it was maybe rye [16:00] scarleo: not fixed yet https://bugs.launchpad.net/ubuntuone-android-files/+bug/952907 [16:00] Ubuntu bug 952907 in Ubuntu One Android Files "Ubuntu One Files hijacks Picasa share option in ICS" [High,Confirmed] [16:00] karni: uh [16:00] but now the upload feature seems broken even after uninstalling u1, I'm 100% sure it worked before [16:01] I'm on a Galaxy SIII now [16:02] I'm on GN, and it never worked for me. I was not able to triage the problem yet :( [16:02] I have no clue how can it influence Picasa. Must be the account manager, _somehow_. [16:02] need to reboot [16:02] I can't understand how one account would influence another. [16:04] karni: maybe u1 creates something that later confuses picasa/media scanner? === gatox is now known as gatox_lunch [16:04] Yes it's very strange. On my last phone, whenever trying to upload to Picasa U1 would pop up and ask for password [16:04] rye: I see my Picasa albums. The problem is with Picasa upload activity, which does not list Picasa albums. [16:05] Lunchtime for me! [16:05] ralsina, have a good lunch :) [16:05] scarleo: thanks :) [16:05] karni: hm, i feel an exception is bound to hide somewhere there [16:05] but i don't have an emulated device at the moment [16:06] too bad Picasa is not open source. I'd rather debug it then think how U1F can possibly influcence anything, even after uninstalling. [16:07] ok, EOD for me, laters! [16:08] if you need something tested just tell me, I'll be here for a couple of hours [16:10] scarleo: thank you. :( http://androidforums.com/samsung-galaxy-nexus/497031-upload-photo-picasa-album-list-blank.html [16:10] Do any of the Ubuntu One Android apps automatically upload photos? [16:11] dotty: Ubuntu One Files [16:11] if configured to do so. [16:11] karni, sweet, thanks [16:11] dotty: You're welcome [16:13] I wish the password requirements for Ubuntu One would be relaxed a touch [16:13] I like to judge my password security myself, not be forced :) [16:13] dotty: To what exactly? [16:14] Just something a little more forgiving [16:14] dotty: Isn't it now 6 chars minimum, and 1 digit? [16:14] * karni is not sure [16:14] Nope.. "Password must be at least 8 characters long, and must contain at least one number and an upper case letter." [16:14] oh. dotty, we're working on relaxing it. [16:14] Oh good :) [16:15] Sounds weird but.. that's a big reason I haven't used U1 much [16:16] Oh man.. [16:16] U1 has been logging photos all along, photos I had since lost :o [16:16] :D [16:16] You are definitely not alone: https://code.google.com/p/android/issues/detail?id=23439 A lot of apps semm to break Gallery->Picasa upload [16:17] so probably Gallery is the broken one [16:18] Hm.. Is there any way to get U1 to download all the photos it has stored at once, rather than clicking on them one by one? === salgado is now known as salgado-lunch [16:21] karni, just removed my Dropbox account and my Gallery->Picasa upload is back [16:21] scarleo: :O ?!? [16:21] scarleo: Also, I'm reading up at http://productforums.google.com/forum/#!msg/mobile/E2tNkWGsALY/SbS5wxUDDfI [16:21] yep [16:22] scarleo: It seems it's an AccountManager bug rather than U1F. That U1F is launched must be a coincidence. [16:22] scarleo: Removed my second gmail account, still not working. Will remove Dropbox account now. [16:23] karni, probbably it's because the email is the same then, so maybe I'm trying to log into u1 with another account [16:24] er.. justhad U1F pop when opening picasa share. kewl. [16:24] still no list [16:25] scarleo: o_O removed my 2nd gmail account, Dropbox account, Ubuntu One account, Picasa is working :O [16:25] scarleo: This is hugeee man, I couldn't sleep well because of that bug! [16:28] hehe, yeah it's scary [16:29] * karni adds db account back [16:30] karni, I tried logging in to u1 with another account so I don't have the same email, upload to picasa works fine :) [16:30] scarleo: WOOOT!?! This is fantastic news [16:31] scarleo: Er.. hehehe... I installed Dropbox, and Picasa is gone from the share menu. [16:31] oh, sorry. it's because it's available in the quick share to the side d'oh [16:31] so somehow the accounts are mixed up due to same email [16:31] And there we go! It's broken again! [16:32] scarleo: You are my hero :) [16:32] :) [16:33] another good reason to use gmail aliases for different accounts ;) [16:33] scarleo: Wow, this is awesome. Thank you! I'm thinking of a possible workaround for folks that want to use their gmail account *and* U1 with same e-mail. [16:33] scarleo: exactly [16:33] The thing is, foo@gmail.com and foo+u1@gmail.com is different for Ubuntu One [16:34] * karni checkes the preferred e-mail setting [16:37] scarleo: Workaround for users: Visit https://login.ubuntu.com/ , Click "Manage email addresses", Add gmail alias (such as youremail+u1@gmail.com), Verify the e-mail, use it to sign in to U1 \o/ [16:37] scarleo: You made my day, I'm serious :) Thank you so much. [16:37] I'll update the bug report. [16:37] Well, thank you for an awesome product :) [16:39] scarleo: I am sooo happy it's not some uber weird bug that U1 account was causing :) [16:39] hehe, me too [16:41] qengho: Awesome news. scarleo here has helped fixing the Picasa hijack bug. Turns out Picasa has clearly a bug, which is causing more than one account in AccountManager with same e-mail address to conflict (be it Dropbox or U1), and it is sufficient to have one of them installed and configured to have Share with Picasa utterly broken (no album listing, no way to upload to Picasa). === gatox_lunch is now known as gatox [16:41] qengho: I already have a workaround (described above) and will udpate the bug report. [16:42] Whoa. [16:42] "fixing" - there's nothing to fix, they should fix Picasa :P [16:42] hehe [16:43] clearly google doesn't want you using anything other than picasa ;) [16:43] hahah! [16:44] scarleo: What's your name? I'd like to thank you in the bug report. Should I use your nickname instead? [16:47] I have a launchpad account: Oscar Tiderman [16:47] scarleo: Thanks Oscar :) [16:48] karni, Glad I could help :) [16:48] scarleo: Please, I want more users like you :D haha [16:49] haha [16:49] (My my reactions here you can really see how badly I felt about this bug. I am so happy haha :) ) [16:51] karni, Yes I can see that, makes me happy too :D [16:52] :)) [16:52] i forgot to mention this, and i just deleted a ton of emails, but if anyone needs reviews today - let me know, it's my day [17:03] Bug #952907 is now Invalid! \o/ [17:03] Launchpad bug 952907 in Ubuntu One Android Files "Ubuntu One Files hijacks Picasa share option in ICS" [High,Invalid] https://launchpad.net/bugs/952907 [17:04] joshuahoover: rye: scarleo: Left a comment (#11) and updated the bug report ↑ [17:11] karni: seriously? Now that's... uhm... [17:11] rye: f'ed up :D [17:11] I am so happy. [17:11] No Critical bugs in U1F now \o/ [17:12] karni: well, picasa upload interface on 2.2 was also not the smartest application (especially with 100500 albums from G+)... but this is way more interesting. [17:12] rye: hahah [17:12] rye: And it was probably picking first account with the default gmail address, which resulted in U1F popping up. d'oh! [17:13] karni: aha, that makes sense now [17:30] briancurtin: saw https://code.launchpad.net/~brian.curtin/ubuntuone-windows-installer/installer-fixes/+merge/111449 [17:30] briancurtin: can I assume this means it now works both for updates and fresh installs? [17:31] ralsina: oh whoops...that is not correct [17:31] i don't know how i pushed that branch [17:31] briancurtin: thought so :-) [17:34] hrmm [17:35] has anyone run the tests for dirspec on osx? [17:36] alecu, ping..... branch updated: https://code.launchpad.net/~diegosarmentero/ubuntuone-client/darwin-fsevents-1/+merge/110382 [17:36] alecu, so, you can review that one, and the windows refactoring too at: https://code.launchpad.net/~diegosarmentero/ubuntuone-client/darwin2-fsevents/+merge/111427 === salgado-lunch is now known as salgado [17:43] gatox: can you pull lp:dirspec on osx and run the tests real quick? [17:43] dobey, yes [17:45] dobey, success [17:47] great, thanks :) [18:31] No home Internet still- ircing from my phone. Not a good day [18:31] mandel, ping.... you still around? [18:41] mmcc :( === zyga is now known as zyga-afk [18:58] ralsina, ping...... 1-1?? if you have time..... if not, no problem [19:04] oh snap [19:04] i am idiot [19:15] dobey: checking your use-dirspec branch... what happened to native_path? [19:16] warnings.warn('native_path will be removed soon', DeprecationWarning) [19:16] so i removed the usage [19:17] dobey: ok then [19:31] Well I might have networking back later this evening if they can figure it out. Meantime I'm buying a laptop [19:33] Anyone here want to discuss moving uniqueapp up a couple of module levels to avoid some unnecessary import order constraints? For background see my great American novel in bug 1015825 and the attached branch [19:33] Launchpad bug 1015825 in Ubuntu One Control Panel "UI does not update after credentials RPC on darwin" [High,In progress] https://launchpad.net/bugs/1015825 [19:35] I'm also curious why the code that worked on windows doesn't work on Darwin. Is there less of a difference between QApplication and QCoreApplication on windows? [19:37] mmcc: no, no problem moving it [19:37] mmcc: it's only used in two places. OTOH, move it where? ;-) [19:38] In the branch I moved it up to ubuntuone.controlpanel.uniqueapp. [19:38] mmcc: we are supposed to keep the Qt-specific stuff segregated [19:39] mmcc: why is importing it first causing that much trouble? Because of imports on the containing __init__.py files? [19:40] The important thing is getting if out from under controlpanel.gui [19:41] Yes it's the init files [19:41] Argh typing on an iPhone!! [19:41] Is slow [19:43] mmcc: understood. Don'tyou have tethering? ;-) [19:43] mmcc: or I can call you on the phone [19:44] Hmmm tethering. No? [19:45] I'm ok typing here, I just like to complain :) [19:45] mmcc: the ideal solution would be for NO import to trigger a reactor installation [19:46] mmcc: since we have been bitten in the past by this [19:47] ideal would be to not have to use qt4reactor. but alas [19:47] dobey: yes, ideal, versus IDEAL ;-) [19:48] Hmmm. There were a few places that triggered it. I went through two trying to see if I could avoid it without moving the uniqueapp init out of main() [19:48] And there were at least three [19:48] Whack a mole [19:48] ralsina: i'm just thinking of the children. :) [19:48] mmcc: we killed a few of those a few months ago, but the little undergrounders breed :-) [19:49] dobey: we'll get rid of it someday, when we don't use twisted for IPC on windows anymore.Maybe. Eventually. [19:49] yeah. and now also on osx [19:50] dobey: at least we are not adding *new* code for IPC that requires twisted ;-) [19:51] Is there a qt ipc solution we could use? Since we use qtnetwork for proxy stuff, qt is in every process we start on macos (iirc) [19:51] mmcc: there is one, but it's unusable for us because we don't use Qt in our lower level layer [19:52] mmcc: ubuntu-sso-client is qt-free for example [19:52] mmcc: and so is syncdaemon [19:54] It is? Qtnetwork is in sso/Utils/webclient , for win & osx [19:55] Line 64 of initpy [19:56] mmcc: right, forgot about that [19:57] Hmm , looking at that file, is-qt4reactor-installed isn't really testing the reactor: :) [19:57] It's testing if a qt app obj exists [19:58] Qt4Reactor implies qapp exists but not the converse [19:58] i don't think there is a way to tell if a particular type of reactor is installed, is there? [19:58] dobey: repr(twisted.reactor).contains("qt") ? [19:58] Yeah I dunno. [19:59] mmcc: looks like Qt doesn't have a portable IPC mechanism [19:59] mmcc: http://doc-snapshot.qt-project.org/4.8/ipc.html [19:59] ralsina: i think doing a repr on twisted.internet.reactor will just tell you it's the module object for twisted.internet.reactor [19:59] dobey: hmmm doesn't that get you the installed reactor object? [20:00] ralsina: i don't see how/why it would, no [20:00] What we really would need is QCOP but it's only available on embedded platforms :-( [20:00] oh, i guess it does [20:01] probably because twisted is doing some pretty nasty stuff with python [20:01] dobey: yeah it does :-) [20:01] Qtnetwork with qlocalsocket would work, no? [20:01] Or the qsharedmemory? [20:02] Which is definitely overkill [20:02] mmcc: and then we need to do a protocol on top of it [20:02] mmcc: we'd have to implement a protocol to send over qlocalsocket i guess; rather than just using one already [20:02] mmcc_: we'd have to implement a protocol to send over qlocalsocket i guess; rather than just using one already [20:04] exactly [20:04] Is there no higher level qt API that we can just instruct to use qlocalsocket? [20:04] so, there *are* protocols we could use, like XMLRPC or JSON-RPC but again, massive headache [20:04] Seems like there should be [20:04] mmcc: looks like it's a hole in Qt's API offerings [20:05] plus, we need it to be async IPC [20:06] Hmmmm [20:06] we could use something like zeromq but... life is short? [20:07] Exactly. I was going to say basically that. Trading known bugs for new ones === mmcc_ is now known as mmcc [20:07] mmcc: so, it's probably "cheaper" to kill a few moles this cycle [20:07] Hm, What about python's multiprocessing? [20:07] mmcc: multiprocessing is sort of evil [20:08] I've used it for toy stuff [20:08] mmcc: it's fork without exec, which means it's easy to get into unsolvable problems [20:08] Ah that's what I was wondering [20:08] * ralsina loves multiprocessing, when used with lots of care [20:09] mmcc: but then you do something like log to a file opened before forking, and *both* processes crash [20:09] Well but do you need to use the process parts? [20:09] Cant you just use Connection objects between processes you created some other way [20:10] I'm just looking for a layer on top of sockets [20:11] No idea, really [20:12] mmcc: http://rpyc.sourceforge.net/ [20:12] Well, probably best to keep up the mole hunt as you said... [20:12] Looking. [20:13] Interesting [20:14] So, twisted is only used for the non-dbus ipc and the tests? [20:15] mmcc: and deferreds all over the place but those are replaceable [20:15] Ah right. [20:15] mmcc: there are some other minor things (there is a process launcher in SSO I think?) [20:16] Yes that rings a bell [20:17] whee [20:18] So back to my fix, since it uses qt, it should not be in the top level package? [20:19] mmcc: right [20:19] putting sso through 2to3 produces a 2300 line diff, but it's mostly really simple stuff [20:19] mmcc: and you get to be the latest to try to fix the "importing things installs reactors" bug [20:19] and with that, i'm heading to the coffee shop, brb [20:19] briancurtin: just 2300? I am strangely optimistic about that! [20:20] ralsina: i'm putting together a wiki page right now with a separation of the dependencies of what's ported, what's not, some twisted stuff, and try to start figuring out estimates [20:20] Ok, I will see if I can shake out the reactor installation bug and avoid moving uniqueapp [20:20] briancurtin, mmcc: cool * 2 [20:21] * ralsina feels so managerial. Manageriastic. Managistral. [20:21] Btw, pretty sure we don't even need uniqueapp on Darwin anyway [20:22] mmcc: we didn't really need it in Ubuntu at first either [20:22] Once packaged, it's hard to get multiple instances started [20:22] Ar first? [20:22] At [20:23] mmcc: you can't launch it twice from the launcher [20:24]