=== zumbi_ is now known as Guest88650 [08:57] Good morning, all! :) [09:01] good morning [09:02] aloha [09:03] hi rye, do you have something new for me? :) [09:04] Moscherkobold: just arrived, let me look at the bug report [09:05] sooo... fix committed [09:07] this means it should work now? [09:15] Moscherkobold: no, the code was not deployed yet and I don't see any merge tickets to merge accounts [09:15] i'll create one and will follow up today (i was off on monday) [09:18] ok, maybe it would be also helpful for me to delete the second account, because it is empty anyway. [09:18] but it is always chosen if i log in on a win pc or on the website [09:19] aquarius: random question, if you have a file shared out and published on U1 and you make changes on the desktop does it automatically get updated and still shared out on the same link? [09:20] czajkowski, it depends how you edit it :( [09:21] aquarius: well I reoponed my presentation and made changes on the deskop and saved [09:21] what other way would you edit it ? [09:21] Moscherkobold: https://bugs.launchpad.net/ubuntuone-servers/+bug/1005819 - if you have a launchpad account you can subscribe to this bug to be notified when we switch it to fix released [09:21] Error: ubuntu bug 1005819 not found [09:22] czajkowski, if the program you edit it with just edits the file, then yes. However, quite a few programs don't *actually* do that (and it's very hard to tell what a program does) -- instead, they save your changes to a temp file and then rename the temp file over the top of the existing file. If this happens, it breaks the public URL because it's actually a different file :( [09:22] czajkowski, and... openoffice is one of the programs that does this rename trick :( [09:22] I would dearly love to see this fixed. [09:22] but fixing it is pretty hard [09:22] aquarius: ahh so that's what happened, had to reshare the presentation as the link died [09:22] czajkowski, yeah. It's really pathologically annoying. :( [09:22] aquarius: it was really handy to do the talk at the weekend, then the file was already on the web and click share. [09:22] aquarius: originally we had a path-based publishing... I recall. then somebody complained and we switched. Was it like this? [09:23] but that kinda bite me last night as needed to make a few changes and add links to stuff I mentioned :/ [09:23] czajkowski, yes. I don't have a very good solution here :( [09:23] rye, path-based publishing fixes this problem, at the cost of others [09:23] aquarius: no worries if there is a bug about it then I'll go back to sipping my drink and being quiet === JamesTai1 is now known as JamesTait [09:37] rye: sry i have no launchpad acc, it would be great if you can keep me up to date? [09:38] Moscherkobold: could you please PM me your e-mail again so that I add it to my notes? [09:38] czajkowski: what I do in this case (i use U1 for all the media on my rtg.in.ua web site) is that i put the files somewhere, edit them and after that copy over to original locations thus overwriting the content byt not changing the inodes [09:39] well, i use a proxy too so that should I break the link i can set up a correct mapping [09:39] but this is an overkill [09:40] yeah I just wanted to add a little content [09:40] no biggie [10:03] Hi everyone [10:03] I just published a zip file from the ubuntuone folder in nautilus [10:04] I can see it is published and see the link on the web U1 interface [10:04] but a right click on the zip file in nautilus still gives me only the "publish" option [10:04] is it a know bug or do I have a problem with my nautilus ? [10:05] "known" [10:06] lalejand: known bug in nautilus plugin which does not load the list of published files properly [10:06] okay thx. Just have to wait for it to be fixed then :) === JanC_ is now known as JanC [11:04] good morning! [11:07] gatox, morning! [11:07] gatox, did you try to run the cp on mac os x? [11:15] mandel, hi [11:15] mandel, no.... i was stuck with a bug in win [11:17] mandel, do you want me to try now? [11:17] * gatox is still trying to reproduce the bug on windows [11:23] gatox, no worries :) [11:23] gatox, just curious [11:56] mandel, gatox: holas! [11:57] alecu, buenas [12:05] alecu, hola! === facundobatista_ is now known as facundobatista [12:28] gatox, I have a new branch for the fsevents project: https://code.launchpad.net/~mandel/ubuntuone-fsevents-daemon/add-dispatcher-tests/+merge/107766 [12:28] gatox, take a look at the description since it tells you how to add OCMock to the $(DEVELOPER_LIBRARY_DIR), specially because the dependency is only related with the tests [12:29] gatox, I tried to write a patch method but although I managed to get it running it was terrible code to mantain, so I went with mocks, so sorry in advanc [12:29] s/advanc/advance [12:38] no brians here [12:38] too [12:41] rye, he starts later :) [12:42] rye, is briancurtin [12:42] ok, lunch time for me [12:43] * mandel lunch [12:44] mandel, ack [12:44] good morning! [12:44] ralsina, hi [12:44] ralsina, once I'm back from lunch, can we take a look on how to setup the mac mini so you can run at least the tests for the objective-c code? [12:44] mandel, did you mark it as private for anything in particular? [12:44] mandel: sure! [12:44] gatox, the entire thing is private [12:44] mandel: I do have a ton of calls today though [12:44] ahhhh ok [12:45] ralsina, once I'm back I'll let you know and then you can ping me whenever you are free :) [12:45] ok, leaving to have tortilla de patatas :) [12:47] mandel, oh, no. No mocks, please! [12:56] hmm [13:00] alecu, the suggested path is what we want! \o/ [13:00] gatox, vaaaamos! [13:01] alecu, so....... fixing! [13:01] alecu, thx for the mumble-time! :P jeje === teknico_ is now known as teknico === m_conley_away is now known as m_conley [13:38] ralsina, alecu when we want to use an icon from the current os theme for u1-cp.... and the os doesn't have that icon.... which would be the proper solution? add the icon to the control panel data folder.... or ignore that icon? [13:38] * mandel back [13:38] gatox: add a fallback to data [13:39] ralsina, roger [13:39] gatox: operative word "fallback" ;-) [13:44] ralsina, let me know when do you have time to set the env for the mac-mini [13:44] ralsina, can be tom if you have a very busy day :) [13:45] mandel: tomorrow early sounds good [13:45] mandel: I want to do all my late 1-1s today [13:46] ralsina, ok, which reminds me, we skipped lasts week [13:46] ralsina, wanna have it or we wait 'til friday? [13:46] mandel: exactly [13:47] ralsina, that means, we have it? [13:48] mandel: yes we will ahve it today [13:49] ralsina, ack [13:55] what is this 1-1 you speak of? [14:05] mandel: i don't quite understand your review of my ubuntuone-installer branch [14:08] dobey, my questions is, if we have a data file (the logo) why don't we copy it to the correct location in the setup.py, or is that logo coming from somewhere else? [14:09] mandel: it is getting picked up/installed automatically [14:09] dobey, the other, please add a test that asserts that we do pass the correct path, I know is a simple 'stupid' tests but it is nice to have [14:09] mandel: i don't understand what you expect to test exactly [14:09] dobey, but what if I get the project and do python setup.py install? the software will work with the logo? [14:09] mandel: yes [14:09] dobey, with the logo you mean, ok cool :) [14:10] dobey: mythical creature [14:10] mandel: debian packages are created by doing setup.py install as well you know :) [14:10] dobey, yes, that is why I'm asking :) [14:11] and the test, a) isn't so simple, and b) makes no sense [14:13] dobey, is simple, self.patch(object, '__find_data_file', function) where function will assert that the passed file is 'u1logo.svg' (which I would store in a const rather than in the middle of the code) [14:13] mandel: and it still makes no sense [14:13] dobey, in makes sense in the way that if someone by mistake removes it or changes it by accident we will know [14:13] asssert that this one string we import in two places, from the same place, is the same [14:13] if it's not the same, your system is f'c [14:13] f'd [14:14] nobody other than me is going to ever touch this code :) [14:14] dobey, well, right now, is not imported from anywhere, is just in that call [14:14] mandel: until it's a const, and it's imported, and then even if someone does change it, the test will still pass [14:14] dobey, and.. while that is very very true (not being touch by any other person) what happens if you do? [14:15] the only way the test will fail is if we hard code it in two separate places [14:15] dobey, well, then place it as a const, and we might find a middle ground ;) [14:15] ubuntuone-instlaler is going to die a very fiery death [14:15] what's the point? [14:15] dobey, self.__find_data_file(U1_LOGO) is much better than self.__find_data_file('u1logo.svg') [14:17] making a constant for a single string is like having a url shortener give you back a longer url than the one you typed in. [14:18] dobey, the same point as all other constants [14:18] dobey, I don't care about the length of the string, I care than when I get told I have to change the logo I don't have to look for it in the code [14:19] mandel: i guarantee that you will never be told to change the logo [14:20] dobey, oh... it is not me who I'm worried about, is the aliens that will find this code as the left overs of our civilization! [14:21] dobey, also, is good practice and is simpler to understand.. pretty please.. [14:21] they will know by the lack of our civilization being around, what idiots we were, already. no need to look at such trifling code [14:22] besides, they probably wouldn't understand it anyway, for thousands of years after its discovery [14:23] also, it wouldn't test that it is the correct logo. [14:23] it would test that we are asking for the filename which we are asking to be loaded, was asked to be loaded [14:24] dobey, we are not longer talking about the tests, I know I've lost that battle [14:24] dobey, I just want you to create the var and used it, no more [14:24] why? [14:24] you hate my fingers? [14:24] dobey, a little ;) [14:27] Good morning, everyone. Good morning, dobey's fingers. [14:28] dobey, is there a problem using an Apache licensed lib with our GPL code? [14:29] possibly [14:29] morning mmcc [14:29] mandel, on a similar testing note... why on earth did you choose OCMock? [14:30] alecu, because there is no bloody simple way to patch methods on objective-c [14:30] my fingers have had enough torture from the ethernet cable yesterday [14:30] dobey, possibly? what does that mean? [14:30] mandel: it means licenses are hard [14:30] mandel: is there anything holding back https://code.launchpad.net/~ballogy/ubuntu-sso-client/drop-lazr/+merge/107153 from being merged? [14:30] mandel, we already had maintenance issues with "mocker-style" tests... [14:30] dobey, yes, that is why I asked you, I usually dont care [14:31] briancurtin, yes, lack of comment [14:31] mandel: i presume you're asking because you want to use an apache licensed library in fsevents-daemon which is GPL? [14:31] mandel, oh, right. briancurtin: It's missing the "commit message", so tarmac won't land it. [14:31] alecu, briancurtin, fixed :) [14:32] mandel, great. [14:32] dobey, yes, I want to use the google-toolbox-for-mac to reuse their logger rather than doing dirty tricks with NSLog [14:32] mandel: cool, thanks [14:33] mandel, what dirty tricks? [14:33] alecu, mandel if you have time for a review please: https://code.launchpad.net/~diegosarmentero/ubuntuone-control-panel/tests-failing/+merge/107799 [14:33] alecu, mandel this fix u1-cp tests and some problems with the music folder [14:33] mandel: http://www.apache.org/licenses/GPL-compatibility.html [14:34] mmcc, well NSLog does not log to a file so you have to change that, it does not support debug only messages which means that i have to use templates to create my own log functions by setting a a DEBUG flag [14:34] mmcc, I would like to log to asl which would be a nice feature [14:34] dobey, thx! [14:35] mandel, re: "no bloody simple way to patch methods on objective-c"... perhaps we should look at what other frameworks are building their tests, and see what they use instead. [14:35] mandel: but please discuss with mmcc re: his question, before going further :) [14:35] dobey, yes [14:35] mandel, and we should look at the plenty other objective-c unit test framework that don't do mocking. [14:36] mandel, like the one that comes standard with xcode. Or the one from google... [14:36] alecu, we are using the one that comes standard with Xcode which is ocunit and recommends ocmock [14:36] mandel, looking at the goog toolbox. I've never really needed a separate logging framework but I'll see what it offers. [14:37] alecu, most common ones are GHunit and ocunit, ocunit is in xcode and that is the only reason why I'm using it [14:37] mandel, also I didn't understand what you meant by using templates...? [14:37] mandel, that sounds like a good reason. Anyway, -1 at using ocmock. [14:37] mmcc, sorry i mean macros.. [14:38] mandel, ah ok. [14:38] alecu, I know, but how do you patch the methods? is possible but is a major PITA ( I have looked at using the objective-c runtime method and method_setImplemenation etc..) [14:38] mandel, I mean... I still need to take a closer look at how ocmock works, but I'm biased by the previous headaches with python mocker. [14:38] alecu, lots of work for what we need [14:39] alecu, yes, that is why I added the comment, I don't like using it, I cannot think of I nicer way to do the tests [14:39] mandel, then probably "patching methods" is not the way we should be writting tests. [14:39] mandel, at least not in objective C. [14:40] alecu, yes, it is not, way to much work to be done and is not our job to write a 'patching lib' [14:40] mandel, I think we should take a closer look at how people are writting tests with ocunit, or ghunit. [14:40] "patching methods" does not work very well in static languages [14:41] dobey, right. Though objective C is pretty dynamic. [14:41] sorry I'm not much help here - my previous objective-C projects were not well tested, I've mostly only used python testing frameworks [14:41] alecu, can you catch me up on the issues you had with python mocker? [14:42] mmcc, the main issue was with maintainability of the tests. [14:42] mmcc: mocking things is bad because it means you end up testing the mocked code, and not the real code; and it makes maintaining the tests much harder [14:42] mmcc, is like writing the tests in brain fuck more or less [14:42] mmcc, a test with mocker would be understandable only to the person writting it, and only for a few hours... [14:43] mmcc, btw: I was responsible for quite a few of those tests, because I liked mocker, and now I'm ashamed. [14:43] heh [14:43] same here.. [14:43] mmcc, so, we are moving out of mocker for new python tests, and replacing it as we refactor old tests. [14:44] ok [14:44] mmcc, also "mocking encourages a bad style of testing that is viral" https://plus.google.com/107994348420168435683/posts/NVc4P3vTZZ9 [14:45] mmcc, mainly using patch from twisted.test case [14:45] mmcc, alecu, the beauty of that is that he wrote mocker.. [14:48] mandel, and that makes him much more trustworthy for recognizing his own errors. right? [14:48] alecu, yes, I was not being sarcastic, it was a complement for the statement :) [14:49] * alecu is shutting down his pristine win 7 vm. This MSDN suscription has payed for itself today! [14:52] alecu, by the way, I remove the kernel header and everything works.. do you have around the errors that you were getting without it? [15:00] me [15:00] me [15:00] me [15:00] mandel alecu mmcc briancurtin gatox thisfred dobey standup in 0 minutes! [15:00] me [15:00] me [15:00] lies! [15:00] me [15:01] alecu: you! [15:01] DONE: bug #1004540, bug #1004486, bug #1003610 TODO: bug #1003610 BLOCKED: no NEXT: gatox [15:01] Launchpad bug 1004540 in U1DB "put docs with revno None (as create doc) and a previously deleted document should work, not require you to know the revno of the deleted version" [Medium,In progress] https://launchpad.net/bugs/1004540 [15:01] Launchpad bug 1004486 in U1DB "db.get_all_docs() that returns all non deleted documents" [Medium,Fix released] https://launchpad.net/bugs/1004486 [15:01] DONE: [15:01] Propose a branch for the tests failing in u1-cp on windows, and the music folder not being shown correctly. [15:01] TODO: [15:01] Fix some other u1-cp and sso issues. Go back to fsevents and add the info obtained with the tests in the wiki. [15:01] BLOCKED: [15:01] Launchpad bug 1003610 in U1DB "get_doc() and get_docs() should not return deleted documents by default" [Medium,Fix released] https://launchpad.net/bugs/1003610 [15:01] No [15:01] briancurtin, go [15:01] DONE: memorial day, BBQ [15:01] TODO: remove lazr from buildout/jenkins/installers once the branch lands, look into creating some type of beta channel so we can hook U1 on windows into the jenkins artifacts [15:01] BLOCKED: no [15:01] NEXT: ralsina [15:01] DONE: mgmt call, tech leads call, proposed a branch for get_root_dir bug, started working on the mac jenkins stuff TODO: keep working on that, lots of 1-1s blOCKED: no, NEXT briancurtin [15:02] next mandel then! [15:02] DONE: bug 1000868 bug 1003519 bug 1003914 bug 1000869 [15:02] TODO: 1-1 with ralsina. Talk with mmcc and alecu, no OCMock, then what? [15:02] BLOCKED: no [15:02] mmcc, please [15:02] Launchpad bug 1000868 in Ubuntu One FsEvents daemon "Missing script to run tests via the command line." [High,In progress] https://launchpad.net/bugs/1000868 [15:02] Launchpad bug 1003519 in Ubuntu One FsEvents daemon "Apple headers are present in the project" [High,In progress] https://launchpad.net/bugs/1003519 [15:02] DONE: visible progress in packaging sso-client, ate hamburgers [15:02] TODO: more packaging fun, reviews [15:02] BLCK: NO [15:02] NEXT: alecu ? [15:02] Launchpad bug 1003914 in Ubuntu One FsEvents daemon "Tests fail when ran from the command line" [High,In progress] https://launchpad.net/bugs/1003914 [15:02] Launchpad bug 1000869 in Ubuntu One FsEvents daemon "Missing tests for the FSEventsDispatcher class" [Medium,In progress] https://launchpad.net/bugs/1000869 [15:02] λ DONE: logo and spacing in installer, reviews, triage, holiday [15:02] λ TODO: expenses, set up milestones, triage [15:02] λ BLCK: none. [15:04] eh, pasto, that should be TODO: DONE: bug #1004673 [15:04] Launchpad bug 1004673 in U1DB "Allow code to pass in document factories" [Undecided,New] https://launchpad.net/bugs/1004673 [15:05] * gatox bank and lunch! === gatox is now known as gatox_lunch [15:10] mandel: wana mumble? [15:10] mmcc, so, about logging, what do you usually do for logging to a file just in debug mode, just a macro? [15:10] ralsina, sure! [15:11] mmcc, I thin asl is a good idea (http://developer.apple.com/library/mac/#documentation/Darwin/Reference/ManPages/man3/asl.3.html) but I don't want to write a logging lib [15:14] uh, I totally forgot about the standup. [15:14] mandel, logging to a file for debug isn't something I've really needed to do - when a program wasn't launched from a terminal, the stdout & stderr go to Console.app, which lets you filter the messages, so I've usually just used that [15:15] DONE: a fix requested for secpatches, techleads mumble, mumble with gatox re suggested paths, started with SD bug queue [15:15] TODO: more SD bug queue, discuss alternative testing framework for osx [15:15] BLOCKED: no [15:17] mandel: but I can definitely see where writing to a separate file, and having configurable log levels, can be nice... [15:18] mmcc, specially for a daemon.. I don't want it to crash and wonder.. [15:18] mandel, crash and wonder what? if it writes to stderr it'll be in the console... [15:19] but I don't mean to argue - writing to a separate file is good, it's just worth seeing if it's enough of an improvement over what's there to be worth bringing in license issues [15:20] mmcc, yes, give me a sec, phone [15:20] mandel - no prob. [15:25] also, why not just use the ASL library as is? looks pretty nice, actually. [15:30] ok, need to get lunch. bbiab [15:31] mmcc, yes, is one of my options [15:33] mmcc, and thinking about it, if adding the google stuff is adding a dep outisde the ecosystem I might as well not use it [15:34] oh, and I'm not on the phone anymore :) [15:34] mandel, yeah. looking at the asl manpage, it looks like it'd be pretty quick to set up.. [15:34] mandel: if you need to log from multiple threads it's a little more work, you need to asl_open a handle on each thread [15:35] but otherwise you can just asl_log(NULL, NULL, ASL_LEVEL_DEBUG, "c-string-msg"); [15:35] mmcc, which we do, but should not be terrible [15:35] and no other setup [15:36] OTOH, I might suggest just going with NSLog and the console until it doesn't do something we need... unless we're already there? [15:36] mmcc, ok, I think I'm convinced to use asl directly [15:36] mmcc, right now we use NSLog, I added the logigng to a file as a bug, but I can leave that bug for later :) [15:37] mmcc, alecu, os for me the only blocking issue is to OCMock or not OCMock [15:37] mandel, sounds good. [15:37] mandel, so the problem is that you want to write tests more like the ones with twisted's patch()? [15:38] mmcc, yes, I want to tests individual methods that interact with other objects and patch/mock those objects [15:38] but ObjC testing is kind of not well developed... [15:38] mmcc, I want to assert the correct interaction but not depend on them [15:39] mmcc, I've seen this in other places, some platforms are behind the TDD movement etc.. for example, 5 years ago in the .Net world nearly no one did unit tests and now they are in the other extreme [15:39] mmcc, as well as with patterns, they did very bad code, now they have as many factories as java :) [15:40] yeah. so, there are ways to replace methods in existing classes, or you could wrap a class and delegate all the methods over to the original object except for the one you want to 'patch'. [15:40] the problem is that there aren't good tools for doing that simply, that I know of [15:41] maybe OCMock can do something like that? but it looks like it's intended to mock whole objects - leading to the problems you guys were talking about earlier [15:42] mmcc, well, you can use stubs which is 'like' a patch /cc alecu [15:42] mmcc, for example [[[event stub] andReturn: [NSArray]] paths] [15:43] mmcc, that will patch the paths method and will always return the same array [15:44] mmcc, well, you have to create the nsarray properly :P [15:44] mandel: so, event is an existing object with a 'paths' method, and this replaces it to always return the argument to andReturn: ? [15:45] that sounds a lot like a patch [15:45] mmcc, yes, exactly that [15:45] mmcc, then you can also do [event expect] which is the same as a patch but with expectations [15:46] mmcc, so if paths is not called, we have an assertion error [15:46] mandel: right, looking at ocmock.org now [15:46] mmcc, so I think it depends on how we use it, we might abuse expect or use a combination of expect and stub correctly [15:47] there's also swizzling, down at the end - which is also basically patching [15:47] mmcc, ha, I did not see that one [15:48] mmcc: swizzling? that's a technical term now? [15:48] ralsina: yes! in objc [15:48] mmcc: omg it is [15:48] sounds like a dance move [15:49] or the process of manufacturing a slurpee [15:49] ah, I see. so with stub, we don't have a real object, and so we might need to mock a lot of behavior even if the test interacts with only a bit of it... [15:49] but with partial mocking and swizzling we can do what patching does - have the original object around, but only replace one method [15:50] aquarius: can you personalise the U1 sharing url in any way ? [15:50] mmcc: that kind of faking you get with stubs is a sort of infectious behaviour, it grows until the complexity doesn't pay of [15:50] czajkowski, not with U1 itself, no. With u1.to you can, though :) [15:51] aquarius: hmm no I was thinking if you're sharing a file, if you personalised the URL then no matter what the update was the url wouldn't change? [15:51] ralsina: right. [15:51] but it looks like we might be able to avoid it and still use ocmock for friendly patching [15:51] mmcc: sounds nice [15:51] czajkowski, ah, right. No. The problem is that a personalised URL would be pointing to a file which is no longer there. (Yes, I know this seems a bit stupid, which is why it's a bug that needs fixing. :)) [15:52] aquarius: no worries was just racking my brain for a work around in case I do it again [15:52] czajkowski: what you need is a URL immortalizer (AKA: a shortener that you can change where it points to) [15:52] czajkowski: I wrote one, never put it in production anywhere, sadly [15:52] ralsina: or I need U1 to note that the file I changed and uploaded is the same file and should keep the same published url. but seeing as I can't I'm asking ye annoying questions that have probably been asked before [15:53] czajkowski: not annying at all :-) [15:53] czajkowski, ya. The problem is that U1 is correctly identifying that they are *not* the same file, but that's not useful -- what U1 should do is cope with that situation even though what it's doing is technically right :) [15:53] technicall V functionality [15:54] no biggie learnt my lesson :) [15:54] czajkowski, exactly, which is why it's a bug ;) [15:54] *technically [15:54] mandel, alecu, ralsina : OCMock's "partial mock" is exactly what I described earlier - a 'wrapper' that forwards methods it doesn't know about to the class it wraps. So if you have something you need to patch, you create a partial mock of it, and use 'stub' to replace the one method - everything else goes to the original object. Sounds like just what we need... [15:54] aquarius: well next time you can't sleep fix zee bug :) [15:54] czajkowski, heh. if it was that easy we'd have fixed it already. Fixing this one is hard. :) [15:54] aquarius: surely that;s a challenge you should accept [15:54] :) [15:55] mmcc: yeah, sounds really nice :-) [15:55] mmcc, that is what I though.. and then use expect for those method you want to assert against [15:55] czajkowski, sure; the question is merely whether fixing this is more important than the other ninety things I'm doing. :P [15:55] aquarius: I'd buy you beer if you fixed this one though! [15:56] you mean you won't buy me beer anyway? :) [15:56] srsly, though, it's surprisingly hard. :( [15:57] aquarius: my kinda beer is baby guinness don't forget that or jaegar bombs! [15:58] mandel, yeah, expect looks useful too - in lieu of blocks, that's the nicest way to patch a method with a replacement that just checks its arguments... [15:59] mmcc, yes, I've done that here: https://code.launchpad.net/~mandel/ubuntuone-fsevents-daemon/add-dispatcher-tests/+merge/107766 line 617 [16:01] mandel oh, I didn't see checkWithBlock: - nice [16:01] mmcc, yes, I think is a really nice API [16:01] mandel: agreed. I think it can be used for good [16:03] so now the fun bit - their license isn't obviously a well-known one. Not GPL, looks bsd-ish. Seems compatible. Is my license pattern matcher broken? https://github.com/erikdoe/ocmock/blob/master/Source/License.txt [16:04] mmcc, and know is when i swear in spanish agains lawers licenses and their mothers.. [16:05] mmcc, seems too, but I'm not sure.. [16:05] mmcc, mandel: we can have ocmock as a build dependency, since it's not something we would be redistributing, right? [16:05] alecu, yes, is just for the tests target, nothing gets shipped [16:05] mandel, then we should not have a license issue with using ocmock. [16:06] mandel, btw: what's "^BOOL" in the checkWithBlocks? [16:06] alecu, and is not even a build dep, you can build the FSEvents target without it, is a test dep [16:06] alecu, is a block, being very rough, a lambda that returns a BOOL [16:06] mandel, what's with the "^" ? [16:07] alecu, block syntax: http://thirdcog.eu/pwcblocks/#objcblocks [16:07] "To define a block variable, the ^ operator is used" [16:08] alecu, and you can name blocks using the same ugly syntax [16:08] * alecu realizes he has done no Objective C code since iOS 2 [16:11] alecu, is not the same the ^BOOL(ConfigMessage *msg) than BOOL(^checkArg)(ConfigMessage *) [16:12] alecu, and one has to be careful with __block and the possible retain of self etc.. is a little messy specially with arc, but does the trick for the tests [16:12] alecu, in the other code I use selectors and performSelector which is nicer is the method is implemented by the class [16:12] * mandel should stop talking about objc [16:13] mandel - no, keep talking until I understand what you just wrote :) [16:13] in which other code are you using performSelector? [16:14] weird...that lazr change seems to fail because ithere are lint warnings for win32 stuff? [16:15] s/warnings/errors [16:15] mmcc, me being super lazy, take a look at line 312 and line 425 of the https://code.launchpad.net/~mandel/ubuntuone-fsevents-daemon/add-dispatcher-tests/+merge/107766 [16:16] mmcc, there I define a dict with the selector name and the message type, then I loop looking for the type and execute the selector, which is nicer than writing a huge if statement [16:16] mmcc, but is me taking advantage of the lang to write as little as possible [16:16] briancurtin, hm, let me check [16:17] briancurtin, fails due to 41: [W0611] Unused import oauth [16:18] I think the rest is garbage from the setup.py build, but I'm not sure [16:18] mandel: which is BS because that change has nothing to do with the branch. i wish that stuff was smarter [16:19] briancurtin, yes, and the major PITA is that we do not own the branch so we have to go after the community guy.. [16:19] mandel: that'd be a pretty easy branch to re-create on our own [16:20] mandel: branch from the branch and use bzr --author [16:21] briancurtin: you too ^ (and here's what I meant: http://blogs.gnome.org/jamesh/2008/05/12/bzr-commit-author/) [16:21] ralsina, I had no idea that could be done! [16:22] mandel: bzr has a bazillion options noone remembers. That is one :-) [16:22] of course I'd want to thank this guy all over the commit messages :) how did he know I would need this for mac packaging! [16:23] mmcc, arch and mac have something in common hehehe [16:23] back === gatox_lunch is now known as gatox [16:23] or was it gentoo, what ever :P [16:24] mmcc: looks like a MIT/X11 thing [16:24] ralsina: the OCMock license? [16:24] mmcc: worse, it's the BSD with attribution license, which is incompatible :-/ [16:24] mmcc: yes, the OCMock license [16:25] mmcc: good news are, we don't need to do derivatives, do we? [16:25] alecu, mmcc, so, do we classify OCMock as evil? Do I find a diff way to run the tests? [16:25] mmcc, is just a test dep nothing for the project [16:25] dobey: does https://github.com/erikdoe/ocmock/blob/master/Source/License.txt look evil to you too? [16:25] ralsina, that was for you ^ [16:25] ralsina: no, we don't need to do derivatives [16:25] mandel, no tool is evil. It's how you use it! [16:25] mandel: if you subclass something from OCMock and ship it with our code, we are shipping a derivative [16:26] mandel: I vote for using OCMock. [16:26] alecu, dammed... that makes anything I use is evil then hehehe [16:26] mandel, so, if you find no better way to "patch" in Objective C, go ahead an use OCMock. [16:26] ralsina: we're not doing that, just a test dep and no subclassing [16:26] mandel, but let's not use it for "mocker-style" tests. [16:26] mmcc: #include'ing or linking? [16:26] mandel: so how should we go forward with this lazr MP? should i jump in and do the fixes with --author? [16:26] ralsina: both, in a separate test bundle that we do not need to ship... [16:26] briancurtin, yes please! [16:27] mmcc: but we do ship everything, because it's open source. [16:27] mmcc: even if we only "ship it" via launchpad [16:27]