=== yofel_ is now known as yofel [09:01] Good morning! === rodrigo__ is now known as rodrigo_ [10:26] Hi all [11:05] hello everyone! [11:13] o/ [11:34] ralsina: ping [11:38] gatox: ping [11:39] nessita, pong [11:39] gatox: hi there! question [11:40] nessita, shoot [11:40] gatox: last week I made a review for you where we skipped a whole testcase because it was using the real control panel. I think you told me you were gonna fix that next, but I'm not sure. Can you please confirm? [11:41] nessita, yes, ralsina told me that he has to implement all that test from scratch [11:41] nessita, it was local_folders [11:42] gatox: what are you working on today? it may be a good exercise for you to do that ;-) [11:42] nessita, i'm working on this: https://bugs.launchpad.net/ubuntuone-windows-installer/+bug/820874 [11:42] gatox: Error: Bug #820874 is private. [11:42] * nessita looks [11:42] gatox: do you have an ETA for that? [11:44] nessita, i think it can be done for 12ART... code complete for sure... tests is another story :P [11:44] gatox: -.- [11:45] gatox: code complete means with tests. Ideally before writing the code... [11:45] gatox: so, when you're asked for an estimates, always always include the tests [11:45] nessita, i mean, i usually spend more time writing tests and fixing when are some changes... [11:45] nessita, ok.... probably it will be ready for lunch [11:46] :P [11:46] gatox: ok, so, can you please take over the fixing tests after completing that task? [11:46] nessita, OK! [11:47] gatox: great! (unless you had something else in your plate... in that case let me know and we evaluate priorities) [11:47] nessita, no, i think that fixing those test should be done before the other bugs that are not so critical [11:48] gatox: GREAT! we're in sync then [12:17] gatox: sso branch approver and globally approved, moving to windows-installer [12:17] nessita, awesome! [12:19] gatox: so, I'm not sure you're adding a test for the removal of the form errors widget. I'm talking this line added in next(): [12:19] self.currentPage().layout().removeWidget(self.form_errors_label) [12:20] nessita, yes, not directly... but that line is executed on "next" for each page that it's called a lot in the tests... but let me add some tests for that [12:23] gatox: yes please, let me explain why is necessary: from the user point of view, is important that the form-error message is cleared, once the error is "resolved" (either by retrying or moving forward) [12:24] nessita, yes, i'm writing some tests tocheck that now every time next() is called in test_gui [12:24] gatox: great [12:24] nessita: i'm sure you asked me for a review, but i've lost the link [12:24] nessita: now would be a good time [12:25] Chipaca: https://code.launchpad.net/~nataliabidart/ubuntuone-client/platform-details/+merge/71723 [12:25] Chipaca: context: we're adding platform details to the ping url in the u1client side. The moving of credentials into its own module is caused by a nightmare of cicurlar imports between code outside platform and code inside there [12:26] nessita, gatox: good morning, and I have that test case redone in this branch: lp:~ralsina/ubuntuone-windows-installer/local-folder-fixes [12:26] ralsina, awesome! :D [12:26] ralsina: hi there! I just made another review like 30 minutes ago, you added changes after that? [12:27] nessita: nope, just sitting down at the computer right now [12:27] nessita: me, I would've also logged the return value of f in log_call [12:27] Chipaca: makes perfect sense, doing it [12:27] ralsina: so, I think you removed the skip of a testcase... I added comments about that in the MP [12:28] nessita: I removed the skip so the tests I added would run, but I can add it back and move them to another tet case. OTOH, in the next branch they would move back [12:28] ralsina: right, the thing is that everything fails for me if that testcase is enabled back [12:28] nessita: also, i would've used urllib.urlencode(dict_of_attrs) rather than assembling it myself [12:28] nessita: so, maybe this branch could go in with the other tests in that testcase failing (but those succeeding) instead of adding code I would remove again later today? [12:29] nessita: (re PLATFORM_PARAMS) [12:29] oh, right it breaks everything for you :-( [12:29] Chipaca: ah, that's a good one, let me fix that as well [12:29] nessita: in fact, looks like all you want is urllib.urlencode(PLATFORM_DATA) [12:30] Chipaca: let me read the doc and I may nod or not :-D [12:30] ralsina: not sure why fixing this branch will make you add code you will remove later today. Can you please help me understand? [12:31] Chipaca: I'm nodding! :-) [12:32] nessita: if I move the tests to another test case and move them back into this testcase again, that means adding the testcase and removing it? [12:32] ralsina: right, but if I'm following you, the test case you will be adding now will be reused in the next branch? [12:33] I mean, you can add the test case you have in the incoming branch (without all the adding of the tests) [12:33] no? [12:33] nessita: actually, since I reimplement most of LocalFoldersPage, the current testcase starts working, AFAIK [12:33] the next branch has this testcase, but working :-) [12:34] ralsina: don't you want to merge that new test case only in this branch? [12:34] new == fixed [12:34] nessita: it will not work for you because of problems in LocalFoldersPage (race conditions) [12:35] I could just merge this branch to the next one and propose the ext one instead [12:35] ralsina: let's do that, and I'll review the next one when is ready (as soon as is ready) [12:35] ok then [12:40] ralsina: any intern work going? [12:40] fagan: just arrived, but I think joshua has some more work on the docs [12:41] joshuahoover: you have anything for me to work on? [12:41] ralsina: well I did a big wiki redirect thing on monday I didnt hear about anything else [12:41] fagan: how about bug triage? :) [12:42] actually that's not a bad idea. [12:42] Ill have a look at the list then :) [12:42] ralsina: fagan completed the last of the big docs tasks we had [12:42] fagan: I need all the NEW/unassigned linux bugs of u1 for the last couple of weeks checked [12:42] ralsina: kk [12:43] joshuahoover: ok [12:45] Chipaca: merged trunk in, urlencode added, test updated, pushed to revno 1111 [12:45] nessita: awesome. [12:45] * Chipaca smiles at beuno [12:46] * beuno giggles [12:46] connect 4 :D [12:48] fagan: ○○○○○○○ [12:48] fagan: ○○○○○○○ [12:48] fagan: ○○○○○○○ [12:48] fagan: ○○○○○○○ [12:48] fagan: ○○○○○○○ [12:49] fagan: ○●●●●○○ [12:49] fagan: sorry, you lose [12:49] * Chipaca runs [12:49] Chipaca: what game were you playing I was white :D [12:50] heh :) [12:50] * fagan is a sharp one [12:50] * Chipaca concedes the match [12:50] gg wp [12:51] nessita: urlencode will produce unordered results, and that test will randomly fail [12:51] uh [12:51] nessita: (because it uses a dict) [12:51] let me patch urlencode then [12:52] nessita: or sort platform_data, or use urlparse to compare :) [12:53] nessita: e.g. urlparse.parse_qsl(urlparse.urlparse(PING_URL).query) [12:53] Chipaca: urlparse will not give me a dict back, unless I'm missing something. I'm getting a ParseResult, where the "path" path looks like a string [12:54] ah, I was using urlparse.urlparse [12:54] yeah, urlprase.urlparse gives you a namedtuple [12:54] the 'query' element therein, you extract with parse_qsl [12:54] ... [12:54] * Chipaca just realized he was reading the code :) [12:55] beuno: I learnt some web programming today! :-D [12:56] nessita: we'll have you doing Plone in no time! [12:56] * nessita fades [13:00] moi [13:00] #nessita { opacity: 0; } [13:01] ralsina, dobey, alecu, nessita, gatox standup [13:01] me [13:01] me [13:02] oh, standup, no notes for me yet, I will go last [13:02] ill go the others can catch up [13:02] DONE [13:02] * a bit of reading [13:02] TODO [13:02] * checking up on recent bugs [13:02] BLOCKED [13:02] * Nope [13:02] gatox: go [13:02] DONE: [13:02] https://code.launchpad.net/~diegosarmentero/ubuntu-sso-client/bug-820872/+merge/71774 and https://code.launchpad.net/~diegosarmentero/ubuntuone-windows-installer/bug-820872/+merge/71773 COMPLETE. Second one finishing some additional tests. [13:02] me [13:02] TODO: [13:03] finish some work in installer UI bugs [13:03] BLOCKED: [13:03] no [13:03] ralsina, go [13:03] DONE: reviews, calls, worked some more on my pending branches. TODO: merge them, hopefully make u1cp do the right thing when there are no credentials BLOCKED: no [13:03] nessita? [13:03] DONE: bug #827343, reviews [13:03] TODO: control panel UI tweaks, more reviews [13:03] BLOCKED: nopes [13:03] NEXT: dobey? alecu? [13:03] Launchpad bug 827343 in ubuntuone-client "Send platform details when pinging for credentials (affects: 1) (heat: 6)" [High,In progress] https://launchpad.net/bugs/827343 [13:04] me [13:04] DONE: chased time.clock issues on VirtualBox; worked on fixing tests for the branch for bug #824819. Started looking at eq_inotify tests [13:04] TODO: fix the f*#@!ng test that's blocking the whole testrun, work on eq_inotify [13:04] BLOCKED: yes, please [13:04] Launchpad bug 824819 in ubuntuone-client "SD and tests need to wait until a watch is removed (affects: 1) (heat: 6)" [High,Triaged] https://launchpad.net/bugs/824819 [13:04] alecu: did upgrading VB helped? [13:05] nessita, no, it does the same. It seems to do it less often though, but it still keeps the clock quiet sometimes [13:06] nessita, the only way I found was to add a small "time.sleep" whenever we are calling time.clock :P [13:06] alecu: can you somehow workaround this locally to unblock? I mean, what concrete symptom are you having? [13:07] nessita, but this is not the root cause of the test that's blocking. The clock issue is a sideturn I took while debugging tritcask tests that use it [13:07] me [13:07] dobey: go! [13:07] λ DONE: bug #827519, discussion on #817133 and #820345 [13:07] λ TODO: finish bug #817133, chat with lisette [13:08] λ BLCK: None. [13:08] Launchpad bug 827519 in ubuntuone-installer "don't add the stable ppa (affects: 1) (heat: 6)" [High,Fix committed] https://launchpad.net/bugs/827519 [13:08] Launchpad bug 817133 in ubuntu "[FFe] [needspackaging] ubuntuone-installer needs packaged (affects: 1) (heat: 12)" [High,Incomplete] https://launchpad.net/bugs/817133 [13:08] alecu: ok, so feel free to "drop" the tritcask issue for now [13:08] nessita, the test that blocks right now just blocks the whole test run when running "run_test.sh" [13:08] alecu: which one is it? [13:09] nessita, but it works fine if I run just that test file [13:09] alecu: is it something in test_vm? [13:09] nessita, no, it's in test_sync [13:09] nessita, most of test_vm is running fine on trunk [13:10] nessita, let me push it, and I'll ask you to run it to see if it hangs like here. [13:10] alecu: sure! [13:11] nessita, lp:~alecu/ubuntuone-client/watch-finished-deferred [13:11] branchinf [13:11] nessita, run_tests -t test_sync works [13:12] nessita, but just "run_tests" hangs up on test_sync.runTest :-( [13:12] nessita, that'll come in hadny for your rotation *wink* [13:12] _ != - [13:13] alecu: FYI: [13:13] Text conflict in tests/platform/test_os_helper.py [13:13] 1 conflicts encountered. [13:13] alecu: can you please fix to ensure we run the same? [13:14] oh, sure [13:16] Chipaca: is there a cleaner way to solve http://pastebin.ubuntu.com/668248/? I don't like using '://' explicit, but the closest thing I found from urlparse is urlunsplit, but to use it in the test I should chopped off the query content from the ParseResult, which is not cleaner nor clearer, from my POV [13:26] nessita, fixed and pushed [13:27] alecu: re-branching! [13:27] nessita, I would not mind about the "://". Just make it a constant, as it would be rare for us to change protocols to something that does not use "://" [13:28] alecu: is a test, and is used in only one place, you think is worth making a constant? [13:28] mailto:ping@url or telnet:ping.url sounds unreasonable [13:28] nessita, then don't mind about it! [13:28] * nessita does not mind then [13:28] Chipaca: fixed and pushed to revno 1112! [13:29] nessita: ok [13:29] nessita: I +1'ed it already [13:29] great! [13:31] alecu: run_tests -t test_sync, all green, going with the whole suite now [13:31] nessita, Chipaca: I would like a review of https://code.launchpad.net/~ralsina/ubuntuone-windows-installer/local-folder-fixes/+merge/71857 if you have a moment please [13:32] ralsina: sure! [13:34] alecu: run-tests finished: FAILED (skips=39, failures=9, errors=42, successes=2141) [13:35] alecu: what test in particular is being blocked for you? [13:35] nessita, test_sync.runTest [13:35] nessita, yes, runTest [13:35] hum, that's a setUp or tearDown being blocked [13:35] nessita, test added https://code.launchpad.net/~diegosarmentero/ubuntuone-windows-installer/bug-820872/+merge/71773 [13:36] alecu: can you add prints to setUp and tearDown to see where the runTest is being blocked? [13:36] nessita, I don't think so. I have put pdb.set_trace() in setup, teardown and an overloaded runtest, and it blocks there. [13:37] nessita, I can try that too [13:37] alecu: so, the runTest is run for every TestCase, which testcase is being blocked? we may wanna check that setUp and his parents, if any [13:41] ralsina: is there a testcase for FolderItem? [13:41] I can't find it myself [13:42] nessita: not explicitly. It's old code, but I will add one. [13:42] ralsina: thanks! [13:44] please make sure to assert that: [13:44] - the checkstate is unchecked [13:44] - the volume id is the one passed as parameter [13:44] - the thread is started if calculate is True, not started if not [13:44] - size is None if calculate, 0 if not [13:44] Also, I would advice moving the self.thread assignment outside the if-else since is the same one no matter the value of calculate [13:45] (I'm adding this to the MP) [13:46] Chipaca, ralsina: are we mumbling today? [13:47] nessita: instead of a testcase, would it be ok if I add tests in LocalFolders testcase? FolderItem is mostly an implementation detail for that class [13:48] ralsina: recommended approach is to have at least a TestCase per each class [13:48] ok [13:48] ralsina: so we can split it out if needed, or if we add more logic we can isolate that [13:48] also, is better for the next one who tries to find the tests [13:49] to either understand who that class is used, or add more tests, et [13:49] c === m_conley_away is now known as m_conley [14:00] alecu: any news? [14:01] nessita, I've checked the parents, and no news. And since when running alone it passes fine, I'm suspecting one of the previous tests is leaving some broken state. [14:02] nessita, so right now I'm deleting one by one the previouses .py testfiles and running run_tests again [14:10] ralsina: I made more comments in the MP, but I paused the review so we can have some checkpoints to avoid confusion about what's fixed and what's pending. [14:10] nessita: ok [14:11] gatox: re-reviewing yours now [14:11] nessita, ok [14:11] gatox: on a qss, shall I set font sizes in px or pts? [14:12] nessita, i do it in px [14:12] gatox: lisseeeeete asked me to set control panel font to 13px, not sure if that is correct [14:12] nessita, yes, i'm using font-size with px [14:12] nessita: about self.assertFalse(self.ui.folders_info is None) I am asserting that because get_info is *not* finished, because it's a deferred. [14:12] ack, thanks [14:13] ralsina: but... folder_info can be None, right? [14:13] you're asserting is not None [14:13] or am I missing something? [14:13] oops, read it the wrong way. get_info is calling a fake deferred that succeeds immediately [14:14] so, shouldn't that always be a value? [14:14] ralsina: answer is "so". Let me explain: [14:14] the bad part is I can't make initializePage be a generator because it's called by PyQt so can't be a generator, so I should probably go the other path you suggest [14:15] ralsina: I think that's the best option, because otherwise, someone who reads that test may think that we ensure that when initializePage ends, folder_Info is not None, which in practice is not True [14:15] ok, makes sense [14:18] gatox: question, were the changes to the ui in the branch before? I think they weren't... [14:18] about the join(): I am making sure the thread is not running because if it is, it will overwrite the item.size later. [14:18] gatox: but please correct me [14:18] * gatox looking... [14:18] ralsina: what if the thread was not running and you join it? can that fail? /me does not know [14:18] nessita: OTOH, that thread should not be started, so removing [14:19] if a thread is not running, join returns immediately [14:19] nessita, nop....... sorry... reverting... [14:19] ralsina: can we start a thread already joined? [14:20] nessita: sure [14:21] yikes, I was not using the "calculate" parameter if it was a UDF, so the thread was started. [14:21] nessita, ralsina: why are we using threads in the Qt part? [14:21] alecu: calculate disk usage of folders in a non-blocking way [14:22] ralsina, do we have the qt reactor running there? [14:23] alecu: yes [14:23] nessita, done [14:23] ralsina, what api are we using to open that thread? [14:23] alecu: python's threading [14:23] alecu: I wish I could use QThread, but we can't [14:23] ralsina, perhaps it would make sense to use the twisted threading [14:24] alecu: big point for using threading: I already wrote it ;-) [14:24] ralsina, my point is that we are not sure if threading conflicts with twisted threading on windows. [14:24] alecu: ugh [14:25] d=reactor.deferToThread(calculate_disk_space) [14:25] d.addCallback(disk_space_returned_something) [14:25] alecu: still ugh because I have the code here and it works :-( [14:25] d.addErrback(disk_space_failed_and_heres_the_exception) [14:25] and it has been there for a month or so [14:25] alecu: good point! [14:26] ralsina, ok, if it works, don't fix it! [14:26] alecu, ralsina: let's file a bug for this and try to schedule this some time in the near future [14:26] alecu: can you please fill the bug? [14:26] fine by me [14:26] nessita, ralsina: I'm pointing this, because I hate joins [14:27] alecu: there are no more joins :-) [14:27] ralsina, cool then. [14:27] alecu: me too, and since ATM very few people is running this, we can't ensure we won t have issues there [14:28] * ralsina points out that rewriting things that are not kown to be broken is very non-TDD [14:29] ralsina: right, but is good to know that we may have weird issues and they may be caused by threads misbehaving in windows [14:29] nessita: sure, we can keep that in mind [14:32] gatox: looks good! though, question: you alse reverted the changes to the qss... I don't think taht's right [14:32] gatox: there was a new style for form errors? [14:33] ugh [14:33] alecu: I'm afraid to ask, but.. yes? [14:33] nessita, oh crap... yes, you are right..... [14:33] * gatox is working with form errors everywhere [14:33] nessita, we are using python threads in syncdaemon too. [14:33] alecu: what for? [14:33] nessita, in hash_queue and the windows NetworkManager [14:34] nessita, I've already asked in #chicharra about the hash_queue, and we'll have to wait for mandel to ask about the windows NetworkManager [14:34] alecu: let's fill bug, there's much we can do right now today [14:34] there isn't much* ;-) [14:34] nessita, done [14:35] oook [14:37] gatox: approved, please seek another review [14:38] nessita, [14:38] ok [14:39] alecu, ralsina can you review this branch? https://code.launchpad.net/~diegosarmentero/ubuntuone-windows-installer/bug-820872/+merge/71773 [14:39] gatox: sure! [14:39] ralsina, thanks! [14:57] nessita: answered comment, pushed fixes [14:57] now let's look at gatox's branch1 [14:58] ralsina: ack, I will re-review in 15 minutes~ [15:00] gatox: shouldn't that fix for critical be done also in sso-client? [15:01] ralsina: is done and merged [15:01] ralsina, yes...... that branch was merge already [15:01] oh, ok, forgot about it [15:01] then it [15:01] 's a+1 [15:01] ralsina, cool [15:02] hmm [15:14] lunch time, bbiab [15:14] ralsina: I just requested the swap day for next MOnday. gatox will do the same, but he's defining which day will swap the holiday to [15:15] nessita: so both of you and mandel will be here on monday? [15:16] ralsina, yes [15:16] ralsina: yeap, gatox needs the holidays for pycon and/or christmas break... [15:16] gatox, nessita: cool, I'll approve later today [15:16] great [15:16] for pycon we can use conference days, can't we? [15:16] ralsina, i think i don't have conference days [15:16] gatox: I am not sure you have any conference days until after the probation though [15:16] ralsina: yes, but seems like gatox has none... [15:17] yes, on probation there are no holidays/conference days [15:17] ralsina, ahhhh, that make sense [15:37] I'm off for a (today) shortish lunch. See you people in about 40 minutes. [15:57] alecu: you still blocked? :-/ [16:03] nessita, I found that by erasing test_filesystem_notifications.py all the tests pass, and now I'm bisecting inside that. [16:04] alecu: the windows version or the multiplatform onw? [16:04] nessita, the windows one [16:05] alecu: let me know if I can help debug [16:05] thanks [16:34] lunchtime for me! [16:35] and back, with belly full === beuno is now known as beuno-lunch [16:57] ralsina: for future coding (no need to change it now), is recommended to call the addCleanup right after the sentence that makes the env "dirty". What I mean is, ideally right after creating a file, for example, we would add the self.addCleanup call [16:57] nessita: ok, I can move it before merging the branch, so I don't make the diff change [16:58] ralsina: one thing I will ask you to fix is that this call: [16:58] self.addCleanup(lambda: shutil.rmtree(self.tmpdir)) [16:58] should be: [16:58] self.addCleanup(shutil.rmtree, self.tmpdir) [16:58] you can fix that later as well [16:58] ok, will do both things together [16:58] thanks [17:05] ralsina, please correct me if i'm wrong... but there is no chance to detect focus in and focus out event except that with eventFilter, isn't it? [17:05] (i', talking about qlineedit) [17:05] gatox: if you set eventTracking, I think there is, let me check [17:06] you want focus events, or mouse hover events? [17:06] focus [17:06] focusInEvent() [17:06] focusOutEvent() [17:07] ralsina: question: I see that the mocked info for volumes_info has 3 items in it, where 2 of them are UDFs; but the test_status_after_get_info checks against 5 top level item count. Which ones are the other 3? [17:07] I think I foooooooooound it.... [17:07] nessita: the default folders (music docs pictures) [17:07] alecu: tell us!!! [17:07] yes...... but i'm using a simple qlineedit inside another widget....... so i should do lineEdit.installEventFilter(self)..... i ask because i know you don't like eventFilter :P [17:07] ralsina: ack [17:07] I think it was a missing "yield" in test_add_single_watch. [17:07] gatox: which means inheriting unless it's an enhancedlineeedit [17:07] I'm testing it all together now. [17:08] gatox: indeed I don't ;-) [17:08] ralsina: is there any test for the adding of those default folders? [17:08] (I want to follow it in the code, if there is any) [17:08] fuck [17:08] alecu: no? :-/ [17:08] nessita: well, it's checked that they are added (or else there would be only 2), but the function that tells you what they are is going away [17:08] no, it was not a missing yield [17:08] ralsina, right..... i was reviewing my options...... because it will be inherit just for focusInEvent and focusOutEvent, and i'll need to replace the components in the UI as promotedWidgets [17:09] ralsina: not sure I understand. What function is going away? default_folders? [17:09] gatox: let me research 5'. If I don't find an alternative, I'll give you a eventFilter exception :-) [17:09] it's an ugly monkeypatch that's not restored :-( [17:09] nessita: yes, it will use the function that's in xdg.basedirectory instead [17:09] alecu: is a call to pacth? [17:09] nessita, no [17:09] Watch.start_watching = start_watching [17:10] guh! [17:10] ralsina: but, the default_folders will remain in place, just the implementation will change, right? [17:10] nessita, it seems mandel only learned about trial.patch pretty recently :-( [17:10] ralsina: what should we do about pyxdg? [17:10] nessita: yes, it will just return three paths from the other function [17:11] ralsina, or i can create an enhanced lineedit..... and maybe include some generic logic inside there to check if they are valid and so..... it's not a problem [17:11] ralsina, what you think is best [17:11] gatox: http://doc.qt.nokia.com/latest/qapplication.html#focusChanged [17:11] so no eventFilter for you ;-) [17:11] ralsina: so we do need a test to confirm that default_folders is returning what it should, not matter what implementation has. Is that test in the suite already, or shall we add a bug for it? [17:12] damn, damn, damn the little childrens [17:12] ralsina, jeje ok [17:12] nessita: it makes more sense to add that test in xdg when it's done, right? [17:12] dobey: good question, didn't have time to do anything about it yet :-( [17:13] nessita, I ended up deleting every test_*.py file except the two involved: test_file_notifications.py and test_sync.py [17:14] nessita, and then commenting the test functions inside there... [17:14] ralsina: so, we need at least 2 (group of) tests: one in u1client when fixing xdg, and another (very likely not the same one) in windows-installer for the default_folders method. This ensures that no matter the implementation of default_folders, we ensure the correctness of the method in installer [17:14] this's been hell :-( [17:14] alecu: I can certainly imagine... :-( [17:14] ralsina: I meant: "no matter the implementation of xdg methods..." [17:15] nessita: not saying not to test the pathsm but there isn't much to check in those paths, since they can be lliterally anything [17:15] ralsina: that's why inside installer we should not depend on anything in particular but we very likely patch xdg with fixed paths and assert over that [17:16] but we need test(s) for default_folders [17:16] nessita: not sure what to test beyond "it's three strings" [17:16] ralsina: I agree we can add them in an incoming branch, but I want to be sure we're in sync [17:16] nessita: no, I will add it I just am not sure what to test for [17:17] ralsina: so, consider this scenario: someone comes and see the default folder implementarion and removes the calls to xdg (or whatever we use) and replace that by the returning of a fixed list with 3 things in it, ['casa', 'perro', 'gato']. We need a test failing because the content of the result, despite having the proper lenght, is not the expected [17:18] ralsina: I know no one will approve such a branch, but is a silly example trying to show what you need to test [17:18] but that is not actually a wrong result (except that they are not absolute paths) [17:18] ralsina: it's if I actually use the string 'casa', 'perro' and 'gato' [17:19] we want to ensure the contents are exactly 'My Documetns' and all, where 'My Documents' is not a literal string but something that we patch inside the underlying libraries [17:19] nessita: you can change your documents path to be "c:\users\casa" (or pretty much any other path) [17:19]