ignore trailing slash in ignore patterns

[Martin Pool]

From: Marius Kruger <email address hidden>

   hi,
   On linux when I do the following with tab completion
   bzr ignore dist/
   it adds a backslash on the end.

   Is it possible to ignore trailing slashes?
   it could be done something like
   p='dist/'
   p=p[:p.rfind(os.sep)]

   OR
   when checking files to ignore
   the matching should match dist/ to dist

   regards
   Marius Kruger

[Wouter van Heyst (larstiq)]

The above snippet will always delete the last character, so it should at least be:

if name_pattern.endswith(os.sep):
    name_pattern = name_pattern.rstrip(os.sep)

But I'm not convinced this is the right approach, one may want to add os.sep for another platform where it isn't (perhaps far fetched). Would it have value to normalize patterns ending on os.sep to end on '/' and check in matching against such a pattern if we are dealing with a directory or not, or would that be too confusing?

[Robert Collins (lifeless)]

I think this is reasonable. internally we normalise all code as being '/' separated, so checking for path[-1] == '/' is sufficient. This should be done in the ui layer - in builtins.py.

[Martin Pool (mbp)]

Ignoring should probably be done when matching ignore patterns, not when taking in the command line arguments. I don't see what "path/" would otherwise mean.

[John A Meinel (jameinel)]

I think we can sanitize both times.
Once when reading patterns, and once when adding them to .bzrignore.

This is still an open bug in 0.9, but is not a blocker. I might look into this as part of my ignore updates.

[John A Meinel (jameinel)]

Kent Gibson wrote a fix for this, which should be available in bzr-0.13