https://pantsbuild.org/ logo
Powered by
#general
  • h

    happy-kitchen-89482

    10/15/2015, 6:42 PM
    plus we end up with complications around src/ vs. src/main
  • h

    happy-kitchen-89482

    10/15/2015, 6:42 PM
    we'd have to be careful to match in the right order, so we don't end up thinking "main" and "test" are languages, in a maven layout situation
  • w

    witty-crayon-22786

    10/15/2015, 6:43 PM
    longest first, yea.
  • w

    witty-crayon-22786

    10/15/2015, 6:43 PM
    think that will always be a thing though.
  • h

    happy-kitchen-89482

    10/15/2015, 6:43 PM
    Not really an issue currently, because we match the language.
  • w

    witty-crayon-22786

    10/15/2015, 6:43 PM
    @happy-kitchen-89482: i agree we want to support common cases, but i don't know about supporting all cases
  • h

    happy-kitchen-89482

    10/15/2015, 6:44 PM
    We have to have some escape hatch
  • w

    witty-crayon-22786

    10/15/2015, 6:44 PM
    @happy-kitchen-89482: it is, because if you allow arbitrary patterns, then someone can define a pattern that includes 
    java
    in it
  • w

    witty-crayon-22786

    10/15/2015, 6:44 PM
    (...it is still the case that longest needs to win)
  • h

    happy-kitchen-89482

    10/15/2015, 6:44 PM
    no one is realistically going to create src/java/python
  • w

    witty-crayon-22786

    10/15/2015, 6:45 PM
    src/py/python ?
  • h

    happy-kitchen-89482

    10/15/2015, 6:45 PM
    but currently the longest one will win, because we don't stop the trie walk until we hit a leaf
  • w

    witty-crayon-22786

    10/15/2015, 6:45 PM
    cool.
  • h

    happy-kitchen-89482

    10/15/2015, 6:45 PM
    but it would be hard to get src/py/python under your scheme
  • w

    witty-crayon-22786

    10/15/2015, 6:45 PM
    alright. i don't feel super strongly about it. just feels like something that's going to break whenever we add a new language.
  • h

    happy-kitchen-89482

    10/15/2015, 6:46 PM
    how frequently do we add new languages?
  • w

    witty-crayon-22786

    10/15/2015, 6:46 PM
    and isn't buying us anything yet. solution in search of problem
  • w

    witty-crayon-22786

    10/15/2015, 6:46 PM
    not often enough!
  • w

    witty-crayon-22786

    10/15/2015, 6:46 PM
    😃
  • h

    happy-kitchen-89482

    10/15/2015, 6:46 PM
    heh
  • h

    happy-kitchen-89482

    10/15/2015, 6:47 PM
    I can probably get rid of the explicit patterns, but I will insist on the fixed patterns as an escape hatch.
  • w

    witty-crayon-22786

    10/15/2015, 6:47 PM
    which 
    options
    are we talking about exactly?
  • w

    witty-crayon-22786

    10/15/2015, 6:47 PM
    so, drop 
    --langs
    ?
  • h

    happy-kitchen-89482

    10/15/2015, 6:47 PM
    I could kill 
    --source-root-patterns
  • h

    happy-kitchen-89482

    10/15/2015, 6:48 PM
    I'm not comfortable with 
    --langs
    but could consider it
  • h

    happy-kitchen-89482

    10/15/2015, 6:48 PM
    We'd need to replace it with some canonicalizing mapping anyway
  • h

    happy-kitchen-89482

    10/15/2015, 6:48 PM
    That information needs to live... somewhere
  • w

    witty-crayon-22786

    10/15/2015, 6:49 PM
    ok. killing --source-root-patterns in favor of an explicit list works for me.
  • h

    happy-kitchen-89482

    10/15/2015, 6:49 PM
    Let me think about capturing the lang from the path
  • w

    witty-crayon-22786

    10/15/2015, 6:49 PM
    i think that if you keep the 
    --lang
    list, it will immediately migrate out when we have first class support for langs
1...550551552...2420Latest