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

    rough-minister-58256

    09/16/2015, 12:14 AM
    also not sure I see a great benefit in ‘double checking’ users intent, tho I’m sure some would disagree. they provided inputs to pants that led it to the publish step, so why annoy them w/ a secondary check?
  • b

    bored-art-40741

    09/16/2015, 12:15 AM
    Today I actually walked someone through a publish and we definitely had to bail on user input
  • r

    rough-minister-58256

    09/16/2015, 12:15 AM
    hah, nice
  • r

    rough-minister-58256

    09/16/2015, 12:15 AM
    under what circumstances?
  • b

    bored-art-40741

    09/16/2015, 12:16 AM
    It's not obvious at all that publish really, really wants to push to upstream programmatically at the end
  • b

    bored-art-40741

    09/16/2015, 12:16 AM
    Perhaps that's the real issue
  • r

    rough-minister-58256

    09/16/2015, 12:16 AM
    hmm
  • b

    bored-art-40741

    09/16/2015, 12:16 AM
    Our general use for publish is, "Take this target and get it on nexus"
  • b

    bored-art-40741

    09/16/2015, 12:17 AM
    The pushdb is handy but sort of secondary, and perhaps not worth the complexity (for us)
  • e

    enough-analyst-54434

    09/16/2015, 12:22 AM
    @rough-minister-58256: because evidence says users don't actually know what they are doing when they click publish
  • e

    enough-analyst-54434

    09/16/2015, 12:23 AM
    This is publishes fault to some degree since it manages versions
  • e

    enough-analyst-54434

    09/16/2015, 12:23 AM
    thus it shows you changelogs customized to each jar you're about to push in the trans graph
  • e

    enough-analyst-54434

    09/16/2015, 12:23 AM
    so you can check a patch-bump is valid
  • e

    enough-analyst-54434

    09/16/2015, 12:23 AM
    in-practice folks do 
    yes| ...
  • r

    rough-minister-58256

    09/16/2015, 12:23 AM
    ah
  • e

    enough-analyst-54434

    09/16/2015, 12:23 AM
    and semver cries
  • r

    rough-minister-58256

    09/16/2015, 12:23 AM
    heh
  • e

    enough-analyst-54434

    09/16/2015, 12:25 AM
    So the startegy may need to be rethought - this goes way back to the any build
  • e

    enough-analyst-54434

    09/16/2015, 12:25 AM
    Basically a hybrid would be nice where you can write down next in the BUILD file
  • e

    enough-analyst-54434

    09/16/2015, 12:25 AM
    for when you do a commit you know should bump a middle bit of the rev
  • e

    enough-analyst-54434

    09/16/2015, 12:26 AM
    but then somehow ignore that field if the time has moved past that?
  • e

    enough-analyst-54434

    09/16/2015, 12:26 AM
    not sure
  • e

    enough-analyst-54434

    09/16/2015, 12:26 AM
  • e

    enough-analyst-54434

    09/16/2015, 12:27 AM
    customized changelogs
  • r

    rough-minister-58256

    09/16/2015, 12:27 AM
    in principle, handling user input in a task just seems to conflict in my mind w/ the idea of tasks as pure functions and potentially running remotely
  • e

    enough-analyst-54434

    09/16/2015, 12:27 AM
    running a remote publish is currently antithetical to publish
  • e

    enough-analyst-54434

    09/16/2015, 12:27 AM
    thus the conflict you see
  • e

    enough-analyst-54434

    09/16/2015, 12:27 AM
    its meant for a human operator right now
  • e

    enough-analyst-54434

    09/16/2015, 12:28 AM
    we may just need to elevate the bit
  • e

    enough-analyst-54434

    09/16/2015, 12:28 AM
    repl, run?, publish
1...371372373...2420Latest