[PATCH 1 of 2 stable] fileset: add tests of generated working copy states

[Martin von Zweigbergk]

On Fri Jan 23 2015 at 4:08:34 AM Yuya Nishihara <yuya at tcha.org> wrote:
>
> I got many diffs with quick workaround for issue4497.
>
> diff --git a/mercurial/cmdutil.py b/mercurial/cmdutil.py
> --- a/mercurial/cmdutil.py
> +++ b/mercurial/cmdutil.py
> @@ -2578,7 +2578,8 @@ def revert(ui, repo, ctx, parents, *pats
>                          return
>                  ui.warn("%s: %s\n" % (m.rel(path), msg))
>
> -            m = scmutil.match(ctx, pats, opts)
> +            # XXX not work with largefiles, can't use 'set:**', etc.
> +            m = scmutil.match(repo[None], pats, opts)
>

That same patch was my first attempt too, but it seems to be a deeper issue
of what context to evaluate the filesets in. Filesets accept a revision so
you can do things like " hg files 'set:added()' -r bead0c7b4f68" and see
which files were added in that revision. However, while most people expect
"hg revert 'set:added()'" to revert added files (i.e. to forget them),
since that is equivalent to "hg revert -r . 'set:added()'" it also makes
sense to say that set:added() should be evaluated in '.'. Perhaps the
answer is, as I think Pierre-Yves suggested, to add a revision parameter to
set:added() and make it default to the working copy, so "hg revert
'set:added()'" still reverts to the parent of the working copy, but the
fileset is evaluated against the working copy. That would mean that "hg
files 'set:added()' -r bead0c7b4f68" instead has to be written "hg files
'set:added(bead0c7b4f68)'". What do you think?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20150123/f32d92a1/attachment.html>