| jelmer | vila: I'm hitting more test failures in sid, looks like this is because pycurl is now stricter | 17:08 |
|---|---|---|
| vila | jelmer: seen that too :-/ but heads down on the ci infra | 17:09 |
| vila | jelmer: I have, I think, a fix for the gettext/lazy_regexp issue though | 17:09 |
| vila | jelmer: I didn't did the pycurl one, stricter for what ? | 17:09 |
| jelmer | vila: basically, when you do proxy requests, they now require that you pass in "http://" URLs for the proxy rather than ignoring the scheme as they did before | 17:12 |
| jelmer | https://github.com/curl/curl/issues/1015 | 17:12 |
| jelmer | so it should hopefulyl just be sufficient to replace http+pycurl with just http before calling pycurl | 17:14 |
| vila | jelmer: the wording in that issue is weird... anyway, having a look as I agree s/+pycurl// may be the thing | 17:15 |
| jelmer | filed https://bugs.launchpad.net/bzr/+bug/1645017 | 17:15 |
| ubot5` | Ubuntu bug 1645017 in Bazaar "proxy tests fail with newer versions of pycurl" [Undecided,New] | 17:15 |
| vila | thanks ! | 17:16 |
| vila | jelmer: hmm, looks like _unqualified_scheme is already used, needs to search if some proxy specific thing is going on | 17:25 |
| jelmer | vila: yeah, I think I found it | 17:42 |
| jelmer | vila: we set 'all_proxy' in the tests to a URL like http+pycurl://joe:bar@127.0.0.1:35245/ | 17:42 |
| jelmer | and then that gets picked up directly by pycurl | 17:42 |
| jelmer | os.environ['all_proxy'] I mean | 17:43 |
| vila | hmm, where ? | 17:43 |
| jelmer | vila: bzrlib.tests.test_http | 17:44 |
| jelmer | vila: it calls self.overrideEnv('all_proxy', ) in many places | 17:46 |
| jelmer | self.overrideEnv('all_proxy', ...) is all over the place :) | 17:46 |
| vila | rhaa, of course, damn it, was climbing the wrong tree. Hmm, but there are ... several tests failing. But yeah, for setting env vars +pycurl shouldn't be there | 17:46 |
| vila | yeah | 17:46 |
| * vila tries hard to page-in the whole context | 17:46 | |
| vila | it's... surprising that this even works, I'm missing some bits, +pycurl should be used to select the client, I'm confused about why/how it's used for the server... (given that the pycurl client will filter +pycurl) | 17:48 |
| vila | or was it just all along that pycurl filtered +pycurl and doesn't anymore in which case ... it's overrideEnv calls that needs to be fixed | 17:49 |
| vila | right, that seems to fix it | 17:52 |
| vila | down to 7 failures | 17:54 |
| fullermd | See, if you'd just gone with perl in the first place... | 17:56 |
| vila | go f... | 17:56 |
| vila | err | 17:56 |
| vila | go for the win ? | 17:57 |
| vila | go fullermd ? | 17:57 |
| vila | ;-) | 17:57 |
| vila | down to 1 failure | 17:57 |
| fullermd | O:-> | 17:57 |
| vila | pycurl.error: (8, '') | 17:57 |
| vila | what on earth is that one ;-) | 17:57 |
| fullermd | Is... that another emoticon? | 17:57 |
| vila | haaaa, indeed, so, probably need to skip that test (they no humor ;) | 17:58 |
| vila | they *have* | 17:58 |
| vila | TestInvalidStatusServer /me sighs | 17:59 |
| vila | jelmer: patch so far: https://pastebin.canonical.com/171950/ | 18:00 |
| vila | jelmer: ftr, this means this is a test-only issue | 18:01 |
| vila | 'E_FTP_WEIRD_SERVER_REPLY': 8, seriously ? | 18:03 |
| fullermd | As FTP goes, that sounds extremely sane. | 18:04 |
| * fullermd is on a campaign to wipe FTP off the face of the 'net. | 18:04 | |
| vila | fullermd: that takes you way too long ! :-p | 18:04 |
| fullermd | Man, you ain't _even_ joking... | 18:05 |
| fullermd | It's amazing. The only real fault of telnet is that it's plaintext, and that was sufficient to practically extinctify it in just a couple years, mostly last millennium. | 18:05 |
| vila | we're dealing with a sabotaged hhtp (ha ha, I'm feeling younger ;) server sending: self.wfile.write("Invalid status line\r\n") | 18:05 |
| fullermd | FTP is plaintext as well, plus horrific eldritch horror of protocol design, and we can't seem to kill the stupid thing. | 18:06 |
| vila | nah, too damn useful to setup insecure servers, let's put it on IOT gateways | 18:06 |
| vila | Ran 764 tests in 2.206s | 18:11 |
| vila | OK (known_failures=4) | 18:11 |
| vila | 89 tests skipped | 18:11 |
| vila | jelmer: with https://pastebin.canonical.com/171952/ | 18:11 |
| jelmer | vila: That's a canonical only pastebin :) | 18:15 |
| vila | vila: you idiot | 18:16 |
| vila | https://paste.ubuntu.com/23538838/ | 18:16 |
| vila | rhaa, didn't press the button on the lp bug... | 18:18 |
| jelmer | vila: thanks! | 18:19 |
| vila | jelmer: can I ask you to fix the debian side while I sort out proper branches for bzr itself (which I'll be able to land when I'm done setting up the ci stuff...) ? | 18:19 |
| jelmer | vila: yeah | 18:20 |
| jelmer | working on it now | 18:20 |
| vila | oh and sorting the mess on the 2.7 and dev branches which still haven't received the post-release cleanups :-/ | 18:20 |
| vila | jelmer: thanks a ton ! | 18:20 |
| fullermd | vila touched it! We all saw him touch it! Now he's on the hook! | 18:20 |
| === vila is now known as wasnt-me | ||
| wasnt-me | who ? | 18:22 |
| === wasnt-me is now known as vila | ||
| jelmer | :) | 18:29 |
| jelmer | vila: new upload on its way | 18:48 |
| vila | jelmer: thanks ! | 19:03 |
| vila | I'll try to followup on the ubuntu side before 2.7.12-7 enter zesty ... | 19:04 |
| jelmer | should be a matter of requesting a sync :) | 19:18 |
| vila | jelmer: right, but that's almost all I know of the process :-/ I still lack practice ... I'm not even sure I know where to ask (but I know who I can ask that question so I should be able to do that ;-) | 19:24 |
| vila | jelmer: upload rejected ? | 22:29 |
| jelmer | vila: yeah, with a weird error. I'm going to try again | 23:12 |
| vila | jelmer: ack | 23:15 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!