[PATCH 3 of 3 V2] lfs: teach the 'lfs()' fileset to handle removed files

[Yuya Nishihara]

On Sat, 10 Feb 2018 22:07:04 -0500, Matt Harbison wrote:
> > # HG changeset patch
> > # User Matt Harbison <matt_harbison at yahoo.com>
> > # Date 1517097753 18000
> > #      Sat Jan 27 19:02:33 2018 -0500
> > # Node ID 8009ea6019880cccef2dd48ea3afad32989f3073
> > # Parent  589f4ea15c31383964e2b7d812bb5431b7dd682e
> > lfs: teach the 'lfs()' fileset to handle removed files

Queued, thanks.

> > The callstatus setting is required to notice the removal of 'lfs.test'  
> > in rev 6
> > in the tests, even though this isn't directly calling mctx.status().   
> > However,
> > it's not needed to get the results in the tests for `hg status`, so I'm  
> > probably
> > missing something.

That's how a subset computed. ctx.walk() wouldn't include removed files.

> In addition to this oddity, I can't get `hg files 'set:lfs()'` to work.   
> The pointerfromctx() method only works for commits, so I special cased  
> `ctx.rev() == None` to just test the name and size against the lfs.track  
> matcher.  (`hg files 'set:lfs()'` seems like a reasonable way to test the  
> config before committing.)

Seems fine.

> The proper files are added to the list  
> comprehension here, but the `[f for f in dmap if match(f)]` in  
> dirstate.matches() fails each match(f).

No idea.