#general
- h
happy-kitchen-89482
11/04/2015, 1:56 AMwe create one per scope... - r
rough-minister-58256
11/04/2015, 1:56 AMMatt has graciously volunteered to write the end-all-be-all python arg parsing lib 😉 - m
mammoth-pharmacist-54215
11/04/2015, 1:56 AMhaha, @rough-minister-58256 likes to put words in my mouth simple smile - h
happy-kitchen-89482
11/04/2015, 1:57 AMThat already exists: http://click.pocoo.org/5/ - h
happy-kitchen-89482
11/04/2015, 1:57 AMIt wasn't suitable for our very specific usage, but it is really awesome - r
rough-minister-58256
11/04/2015, 1:57 AMyeah, a few folks here seem to like that - but imo it still has limitations - m
mammoth-pharmacist-54215
11/04/2015, 1:57 AMneed to head out for an appointment. - m
mammoth-pharmacist-54215
11/04/2015, 1:58 AM@happy-kitchen-89482: let me know if you are interested in syncing up at some point. - h
happy-kitchen-89482
11/04/2015, 1:59 AMsyncing up about...? - r
rough-minister-58256
11/04/2015, 2:02 AMMatt took off, but I think he meant syncing up about the idea of writing a next-gen arg parsing lib that could be a drop-in for pants. - r
rough-minister-58256
11/04/2015, 2:03 AMthere are at least a couple of bigger projects that could benefit from a well designed arg parsing lib - e.g. one that handles hierarchical scopes, inheritance, etc ootb - r
rough-minister-58256
11/04/2015, 2:06 AMI wonder if we expect the current options system to remain largely intact post-engine refactor? - w
witty-crayon-22786
11/04/2015, 2:07 AMdon't see why it wouldn't. - r
rough-minister-58256
11/04/2015, 2:07 AMcool - s
salmon-summer-38098
11/04/2015, 2:08 AMthose two things are somewhat unrelated though. what you have been proposing Kris is independent of the engine refactor - s
salmon-summer-38098
11/04/2015, 2:09 AMI still think its a good idea and should be circulated with the pants community -- I think when you and Matt have the prelim design for it, it would be good to share - w
witty-crayon-22786
11/04/2015, 2:10 AMalso, what already exists in pants is 90% of "scopes, inheritance, etc" - w
witty-crayon-22786
11/04/2015, 2:10 AMso if anything, a simpler arg parser might be possible. - r
rough-minister-58256
11/04/2015, 2:11 AMyeah, would largely be a ‘start from what pants has, generalize and work backwards' - r
rough-minister-58256
11/04/2015, 2:11 AMMatt has a doc started - he’s taking point on this. I’m just a passive observer - w
witty-crayon-22786
11/04/2015, 2:12 AMhm - r
rough-minister-58256
11/04/2015, 2:12 AMone of the bigger goals that I’ve heard tho is around performance - nick’s idea of compiling options into something that can be validated more quickly, etc. part of getting us to go-speed. - w
witty-crayon-22786
11/04/2015, 2:13 AMor just have the daemon do arg parsing and use pypy - r
rough-minister-58256
11/04/2015, 2:14 AMuntil we package up our own interpreters and ship them with pants, we should probably be optimizing for the cpython case - r
rough-minister-58256
11/04/2015, 2:14 AMand arg parsing as-is today is an inline action - even in the daemon, has to be done before anything real can fire - r
rough-minister-58256
11/04/2015, 2:15 AMtho the compiled case data could be held in-daemon - w
witty-crayon-22786
11/04/2015, 2:15 AMeither way. that would let you "register_options" for all tasks just once. don't think the parsing itself is that expensive. - r
rough-minister-58256
11/04/2015, 2:16 AMah sure - r
rough-minister-58256
11/04/2015, 2:18 AMI think that’s also the idea behind ‘compiling options’ - you register_options -> some intermediate data structure thats more easily checked, but would make sense in-daemon too - r
rough-minister-58256
11/04/2015, 2:19 AMat any rate, sideproject material. bigger 🐟 to fry