IRC logs for #baserock for Tuesday, 2017-03-28

*** jude_ has joined #baserock07:03
*** ctbruce has joined #baserock07:35
*** jude_ has quit IRC07:48
*** jude_ has joined #baserock07:48
*** rdale has joined #baserock07:49
*** locallycompact has joined #baserock07:56
*** jonathanmaw has joined #baserock08:21
*** gtristan has joined #baserock08:45
*** toscalix has joined #baserock08:54
*** ssam2 has joined #baserock09:14
*** ChanServ sets mode: +v ssam209:14
*** rdale has quit IRC09:35
*** rdale_ct has joined #baserock09:36
*** locallycompact has quit IRC09:41
*** jjardon has quit IRC10:04
*** locallycompact has joined #baserock10:08
*** locallycompact has joined #baserock10:09
*** jjardon has joined #baserock10:15
*** locallycompact has quit IRC10:27
*** locallycompact has joined #baserock10:27
*** toscalix has quit IRC10:54
*** bjdooks_ has quit IRC11:42
*** bjdooks has joined #baserock11:42
paulsherwoodrjek: i'm trying to address https://gitlab.com/baserock/ybd/merge_requests/324 without using wget12:50
paulsherwoodso trying to use the requests library12:51
paulsherwoodunfortunately i'm starting to think that going that route will lead to forcing requirements for later python and other things...12:53
paulsherwoodso i'm starting to think that wget is 'good enough' since ybd already uses it12:53
paulsherwooddo you have any thoughts on this?12:54
SotKooi what is wrong with needing later python?12:55
SotKor specifically, what version of python needs supporting that is older than required by requests?12:57
* paulsherwood wishes there was a standard word/phrase to cover 'as far as possible, don't force users to upgrade their software or install more dependencies just to run your stuff'12:58
*** gtristan has quit IRC13:00
paulsherwoodSotK: in another channel, i hit a bug i when trying with the requests library on an authorised site, and was told 'your python is prehistoric - it does not support Server Name Indication, which is essential for accessing virtually-hosted TLS sites'13:00
paulsherwoodforcing the user to upgrade python and/or requests seems worse than insisting on uptodate wget, to me13:01
rjekuse requests if the user's Python is recent enough, fall back to wget if not13:01
rjekwget will be more fragile13:02
paulsherwoodi guess i can try both in that order, irrespective of python version13:02
rjekI would suggest checking Python version, and if requests fails when it should work, abort.13:03
paulsherwoodwhy is that better?13:03
rjekOtherwise requests may be trying to communicate a security issue to you, which then gets ignored/swallowed/misreported by wget13:03
* paulsherwood doesn't like this very much13:04
* rjek prefers to not ignore security-related diagnostics :)13:05
SotKhttp://docs.python-requests.org/en/master/community/faq/#what-are-hostname-doesn-t-match-errors suggests it might be possible to work around ancient python, but it'd mean installing more things13:05
rjekYeah, I pointed at that earlier in another place13:06
rjekybd does require a load of modules being installed IME, perhaps adding PyOpenSSL is not too onerous13:07
ssam2as long as it doesn't embed a copy of OpenSSL which it tries to build with a random C compiler from the host when installed with `pip`13:08
rjekhah, yes13:08
paulsherwoodrjek: 'wget will be more fragile' - can you elaborate on that?13:09
* paulsherwood is still tempted to just stick with wget...13:09
rjekpaulsherwood: You'll have to parse the output of wget to decide what the error was programmatically, the escaping rules when shelling out are complex for URLs and credentials, different versions of wget may have different properties, wget may not even be installed (it's not standard)13:10
rjekIt's probably very sensitive to the environment in terms of what it does too13:10
paulsherwoodwget has to be installed... ybd already uses it. the other errors i think i'm ok with13:11
paulsherwood(since if a download fails, the user will have to fudge with either wget or their definitions)13:11
rjekOne could say if you remove the dependancy on wget and add the dependancy for PyOpenSSL or something then you still have a net zero change in faff13:11
paulsherwood(or put the required binary somewhere else13:12
rjekAll I'm saying is that is not what I would do.13:12
paulsherwoodack13:12
paulsherwoodthere are many things about ybd that are not something anyone else would do :)13:12
paulsherwoodand i'm not claiming that's a good thing :-)13:13
rjekSNI should work on any Python from after 2014-12-1013:15
rjek(Out of the box, 2.7.9 onwards ship with the support rather than you having to add it)13:15
rjekEven current Debian has that, which is why I describe anything prior to 2.7.9 as pre-historic :D13:16
SotKubuntu trusty has 2.7.5 iirc13:18
*** locallycompact has quit IRC13:25
*** locallycompact has joined #baserock13:25
*** locallycompact has quit IRC13:51
*** edcragg has quit IRC13:51
*** CTtpollard has quit IRC13:51
*** jonathanmaw has quit IRC13:51
*** tiagogomes has quit IRC13:51
*** ssam2 has quit IRC13:51
*** ctbruce has quit IRC13:51
*** CTtpollard has joined #baserock13:54
*** tiagogomes has joined #baserock13:54
*** ssam2 has joined #baserock13:54
*** ChanServ sets mode: +v ssam213:54
*** ctbruce has joined #baserock13:54
*** paulwaters_ has joined #baserock13:55
*** locallycompact has joined #baserock13:55
*** jonathanmaw has joined #baserock13:57
*** gtristan has joined #baserock13:58
*** edcragg has joined #baserock13:58
*** ctbruce has quit IRC15:20
*** jonathanmaw has quit IRC17:03
*** locallycompact has quit IRC17:12
*** jude_ has quit IRC17:36
*** ssam2 has quit IRC18:19
*** gtristan has quit IRC18:55
*** jude_ has joined #baserock19:28
*** locallycompact has joined #baserock20:39
*** locallycompact has quit IRC20:55
*** tiagogomes has quit IRC21:43
*** jude_ has quit IRC22:33
*** jude_ has joined #baserock22:34
*** jude_ has quit IRC22:39

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!