D2988: fix: use a portable python script instead of sed in test

[mharbison72 (Matt Harbison)]

mharbison72 added a subscriber: durin42.
mharbison72 added a comment.


  I'm not sure what the state of this is (`hg phabread` failed for 3022 and 3023), but this works on Windows.  I've got a separate patch to fix the printfs (which doesn't help the {rootpath} issue I mentioned).
  
  In https://phab.mercurial-scm.org/D2988#48851, @hooper wrote:
  
  > > - subprocess: /usr/bin/python $TESTTMP/uppercase.py 1-1 3-3 +  subprocess: c:/Python27/python.exe $TESTTMP/uppercase.py 1-1 3-3
  >
  > I had wondered if there's a reason we don't substitute $PYTHON like we do $TESTTMP? The glob makes the test a little weaker.
  
  
  Not sure, other than the invocation via $PYTHON is relatively new.  I think @durin42 was the one who did most of that work.  Masking the output seems useful too.  (Hopefully it doesn't cause a headache on Windows.)
  
  >> I'm not sure if you have any idea why, but it seems like {rootpath} is getting lost on Windows.  (The changes here are because the printfs don't get run properly on Windows when run directly.)
  > 
  > Not sure.

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D2988

To: hooper, #hg-reviewers
Cc: durin42, mharbison72, mercurial-devel