[PATCH] stripping trailing slashes from ignore patterns (#4559)

[Aaron Bentley]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Kent Gibson wrote:
> Latest changes incorporating John Arbash Meinel's review comments for
> Cheuksan Edward Wang's ignore_multiple_3.patch (phew - say that 3
> times).  This diff applies to that patch and fixes Cheuksan's few
> remaining problems as well as the corresponding problems in my patch.
> Cheuksan - sorry for the pre-emptive patch, but there is quite a bit of
> overlap.
> 
> Used a diff format this time instead of a bundle.  Is there any
> preference? The bundles seem to grow rather dramatically.

We like to preserve metadata, and bundles let us do that without needing
access to your branch.  So if you're providing a diff rather than a
bundle, we like to also have a branch location that we can merge from.

In fact, right now, I'm not sure how to apply this patch in order to
review it or merge it.  So a branch would be handy.

It seems like in the long term, bzr ignore should be able to handle
*files*, not just patterns, and then it would make sense to accept
absolute paths, (like 'bzr add' does).  But this makes sense for now.

> === modified file 'bzrlib/ignores.py'
> --- bzrlib/ignores.py	2006-07-31 14:04:26 +0000
> +++ bzrlib/ignores.py	2006-10-11 14:44:08 +0000
> @@ -99,7 +99,7 @@
>      """Read in all of the lines in the file and turn it into an ignore list"""
>      ignored = set()
>      for line in f.read().decode('utf8').split('\n'):
> -        line = line.rstrip('\r\n')
> +        line = line.rstrip('/\r\n')
>          if not line or line.startswith('#'):
>              continue
>          ignored.add(line)
> @@ -163,6 +163,7 @@
>      ignored = get_user_ignores()
>      to_add = []
>      for ignore in new_ignores:
> +        ignore = ignore.rstrip('/')
>          if ignore not in ignored:
>              ignored.add(ignore)
>              to_add.append(ignore)

It seems like we're stripping slashes in an awful lot of places.  Since
get_user_ignores uses parse_ignore_file, isn't this redundant?


> +    def test_ignore_no_arguments(self):
> +        """'ignore' with no arguments returns an error"""
> +        self.runbzr('init')
> +        out, err = self.runbzr('ignore', retcode=3)
> +        self.assertEqual(
> +            'bzr: ERROR: ignore requires at least one '
> +            'NAME_PATTERN or --old-default-rules\n', err)
> +        self.assertEqual('', out)

^^^ not sure whether you're aware, but you can also do

run_bzr_error(('bzr: ERROR: ignore requires at least one '
               'NAME_PATTERN or --old-default-rules\n',), 'ignore')

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFFLW8j0F+nu1YWqI0RAr2EAJ9eif3aEAB9gI06+FXP5KbhoJY/hwCdGAce
bqXyDl0DkTcD3+k1b3f8NAo=
=I8UQ
-----END PGP SIGNATURE-----