https://pantsbuild.org/ logo
Powered by
#development
  • w

    witty-crayon-22786

    04/25/2017, 11:18 PM
    to the tune of steve ballmer: "declarative! declarative! declarative!"
  • w

    witty-crayon-22786

    04/25/2017, 11:18 PM
    yea.
  • r

    rough-minister-58256

    04/25/2017, 11:18 PM
    what would 
    @classmethod def fields
    entail/return?
  • w

    witty-crayon-22786

    04/25/2017, 11:19 PM
    return [SourcesField]
    for a simple usecase
  • r

    rough-minister-58256

    04/25/2017, 11:19 PM
    ah
  • r

    rough-minister-58256

    04/25/2017, 11:19 PM
    typed fields
  • r

    rough-minister-58256

    04/25/2017, 11:19 PM
    mhmm
  • f

    fancy-queen-20734

    04/25/2017, 11:20 PM
    We’d probably want an ordered list of (name, type) entries
  • f

    fancy-queen-20734

    04/25/2017, 11:20 PM
    definitely with some special descriptioning
  • r

    rough-minister-58256

    04/25/2017, 11:20 PM
    makes sense - was thinking of e.g. 
    Payload
    as a mapping, but yeah can do that entirely with type checks
  • w

    witty-crayon-22786

    04/25/2017, 11:20 PM
    and then the optional "factory"/imperative/etc bit would take the kwargs and the default payload fields or something.
  • w

    witty-crayon-22786

    04/25/2017, 11:21 PM
    ...actually, hm.
  • w

    witty-crayon-22786

    04/25/2017, 11:21 PM
    ...so in the completed model, all of the arg parsing could happen via Fields
  • f

    fancy-queen-20734

    04/25/2017, 11:22 PM
    we’d need to work out the effect on docs too, but I think this model would make docs easier in some ways
  • w

    witty-crayon-22786

    04/25/2017, 11:22 PM
    so the imperative bit would not need the kwargs, per-se... it would basically just generate a complete dependency list given the payload
  • w

    witty-crayon-22786

    04/25/2017, 11:22 PM
    Copy code
    @classmethod
def compute_dependencies(cls, payload_fields):
  return payload_fields['dependencies'] + payload_fields['other-field-with-deps']
  • r

    rough-minister-58256

    04/25/2017, 11:23 PM
    hmmm
  • w

    witty-crayon-22786

    04/25/2017, 11:24 PM
    where "complete dependency list" is a concept that doesn't make sense in the very long term
  • w

    witty-crayon-22786

    04/25/2017, 11:24 PM
    but is necessary for now
  • w

    witty-crayon-22786

    04/25/2017, 11:25 PM
    ... but then, realistically, is there anything we can do in that direction this week?
  • w

    witty-crayon-22786

    04/25/2017, 11:25 PM
    ...or the next two i guess.
  • f

    fancy-queen-20734

    04/25/2017, 11:25 PM
    Yeah, that feels like a multiweek thing.
  • f

    fancy-queen-20734

    04/25/2017, 11:25 PM
    I think we could come up with a design though
  • f

    fancy-queen-20734

    04/25/2017, 11:26 PM
    and a proof of concept
  • r

    rough-minister-58256

    04/25/2017, 11:26 PM
    technically we only need to get this working for e.g. a 
    DependenciesField
    this week to replace traversable_dep_specs
  • r

    rough-minister-58256

    04/25/2017, 11:27 PM
    then potentially iterate further
  • w

    witty-crayon-22786

    04/25/2017, 11:27 PM
    ...and any other fields used in 
    traversable_dep_specs
  • w

    witty-crayon-22786

    04/25/2017, 11:27 PM
    right?
  • r

    rough-minister-58256

    04/25/2017, 11:27 PM
    maybe not, with imperative shims?
  • w

    witty-crayon-22786

    04/25/2017, 11:27 PM
    but yea, that might be a smaller set
1...320321322...906Latest