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

    enough-analyst-54434

    09/04/2015, 11:52 PM
    per my example above
  • w

    witty-crayon-22786

    09/04/2015, 11:52 PM
    see the added test
  • w

    witty-crayon-22786

    09/04/2015, 11:53 PM
    or maybe i'm missing something =/
  • e

    enough-analyst-54434

    09/04/2015, 11:53 PM
    I actually added my example to that test
  • w

    witty-crayon-22786

    09/04/2015, 11:53 PM
    do you agree about 
    os.path.join
    though?
  • e

    enough-analyst-54434

    09/04/2015, 11:53 PM
    Basically fast_relpath does not act like os.path.relpath for the case:
    Copy code
    self.assertEqual('b/', fast_relpath('/c/b/', '/c/'))
self.assertEqual('b', os.path.relpath('/c/b/', '/c/'))
  • e

    enough-analyst-54434

    09/04/2015, 11:54 PM
    I agree os.path.join works fine
  • e

    enough-analyst-54434

    09/04/2015, 11:54 PM
    I think adding that 3rd? fix so fast_relpath works like os.path.relpath would be a good bonus
  • e

    enough-analyst-54434

    09/04/2015, 11:54 PM
    but will not block obviously, up to you
  • w

    witty-crayon-22786

    09/04/2015, 11:55 PM
    there are plenty of ways in which it doesn't act like os.path.relpath, because it does allll kinds of stuff
  • e

    enough-analyst-54434

    09/04/2015, 11:55 PM
    k
  • e

    enough-analyst-54434

    09/04/2015, 11:56 PM
    if it doesn't work like os.path.relpath it probably should just be inlined where its used
  • e

    enough-analyst-54434

    09/04/2015, 11:57 PM
    or have excellent thourough pydoc to set exact expecatations
  • w

    witty-crayon-22786

    09/04/2015, 11:57 PM
    @enough-analyst-54434: hm, ok. but it asserts its expectations
  • e

    enough-analyst-54434

    09/04/2015, 11:58 PM
    except for this case!
  • e

    enough-analyst-54434

    09/04/2015, 11:58 PM
    the user will expect rightly it behaves the same if it does not assert
  • w

    witty-crayon-22786

    09/04/2015, 11:58 PM
    i believe the pydoc says it doesn't normalize
  • w

    witty-crayon-22786

    09/04/2015, 11:59 PM
    but sure, we can inline it there if you'd like.
  • e

    enough-analyst-54434

    09/04/2015, 11:59 PM
    k - well this bug is proof of the issue
  • w

    witty-crayon-22786

    09/04/2015, 11:59 PM
    ...no, it isn't
  • w

    witty-crayon-22786

    09/04/2015, 11:59 PM
    that was not the bug here.
  • e

    enough-analyst-54434

    09/04/2015, 11:59 PM
    none of our brain compilers were smart enough
  • w

    witty-crayon-22786

    09/04/2015, 11:59 PM
    @enough-analyst-54434: that didn't cause the bug.
  • e

    enough-analyst-54434

    09/04/2015, 11:59 PM
    I know - there were 3 layers
  • e

    enough-analyst-54434

    09/05/2015, 12:00 AM
    however making a fast version of an api that produces different results for the cases it accepts is a trap
  • e

    enough-analyst-54434

    09/05/2015, 12:01 AM
    unless the pydic is exhaustive, like it or not, os.path is what people will and should expect
  • w

    witty-crayon-22786

    09/05/2015, 12:02 AM
    @enough-analyst-54434: which trap are we talking about here?
  • e

    enough-analyst-54434

    09/05/2015, 12:02 AM
    util
  • e

    enough-analyst-54434

    09/05/2015, 12:02 AM
    its sold as generic for concomitant trees
  • e

    enough-analyst-54434

    09/05/2015, 12:02 AM
    but it acts deifferent than the standard util
1...298299300...2420Latest