[PATCH V2] revset: make id() to not abort on an unknown 40-byte string

[Martin von Zweigbergk]

On Mon, Apr 13, 2015 at 10:55 PM Ryan McElroy <rm at fb.com> wrote:

> On 4/13/2015 10:40 PM, Alexander Drozdov wrote:
> > # HG changeset patch
> > # User Alexander Drozdov <al.drozdov at gmail.com>
> > # Date 1428988858 -10800
> > #      Tue Apr 14 08:20:58 2015 +0300
> > # Node ID 31e872ad0473ed27cf48b518fee28f34049a9698
> > # Parent  52ff737c63d2b2cb41185549aa9c35bc47317032
> > revset: make id() to not abort on an unknown 40-byte string
> >
> > Instead, just return an empty set in the case as for less-then-40-byte
> > strings.
> >
> > diff -r 52ff737c63d2 -r 31e872ad0473 mercurial/revset.py
> > --- a/mercurial/revset.py     Thu Apr 09 16:18:38 2015 -0400
> > +++ b/mercurial/revset.py     Tue Apr 14 08:20:58 2015 +0300
> > @@ -1293,7 +1293,10 @@
> >       # i18n: "id" is a keyword
> >       n = getstring(l[0], _("id requires a string"))
> >       if len(n) == 40:
> > -        rn = repo[n].rev()
> > +        try:
> > +            rn = repo[n].rev()
> > +        except error.RepoLookupError:
> > +            rn = None
> >       else:
> >           rn = None
> >           pm = repo.changelog._partialmatch(n)
> > diff -r 52ff737c63d2 -r 31e872ad0473 tests/test-revset.t
> > --- a/tests/test-revset.t     Thu Apr 09 16:18:38 2015 -0400
> > +++ b/tests/test-revset.t     Tue Apr 14 08:20:58 2015 +0300
> > @@ -554,6 +554,21 @@
> >     hg: parse error: rev expects a number
> >     [255]
> >
> > +Test hexadecimal revision
> > +  $ log 'id(2)'
> > +  abort: 00changelog.i at 2: ambiguous identifier!
> > +  [255]
> > +  $ log 'id(23268)'
> > +  4
> > +  $ log 'id(2785f51eece)'
> > +  0
> > +  $ log 'id(d5d0dcbdc4d9ff5dbb2d336f32f0bb561c1a532c)'
> > +  8
> > +  $ log 'id(d5d0dcbdc4a)'
> > +  $ log 'id(d5d0dcbdc4w)'
> > +  $ log 'id(d5d0dcbdc4d9ff5dbb2d336f32f0bb561c1a532d)'
> > +  $ log 'id(d5d0dcbdc4d9ff5dbb2d336f32f0bb561c1a532q)'
> > +
> >   Test null revision
> >     $ log '(null)'
> >     -1
> >
> This is definitely one of the cases where I'd love to see the
> before-to-after behavior diff in the test. That being said, this change
> seems fine to me.
>
> It still seems a little weird that an ambiguous identifier is a hard
> error (abort [255]) but a totally missing identifier is not an error at
> all, but I'm not sure if there's anything we can or should do about that.
>

I agree that it is a little weird, but I also don't know what to do about
it. I think hidden revs get reported before this change but I think they're
silent after. That also seems weird, but at least consistently weird.


> _______________________________________________
> Mercurial-devel mailing list
> Mercurial-devel at selenic.com
> http://selenic.com/mailman/listinfo/mercurial-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20150414/bd48518a/attachment.html>