[PATCH] dagop: raise ProgrammingError if stopdepth<=0

[Martin von Zweigbergk]

On Fri, Jun 23, 2017 at 11:12 PM, Yuya Nishihara <yuya at tcha.org> wrote:
> On Fri, 23 Jun 2017 23:06:19 -0700, Martin von Zweigbergk wrote:
>> On Fri, Jun 23, 2017 at 11:03 PM, Yuya Nishihara <yuya at tcha.org> wrote:
>> > On Fri, 23 Jun 2017 22:48:23 -0700, Martin von Zweigbergk via Mercurial-devel wrote:
>> >> # HG changeset patch
>> >> # User Martin von Zweigbergk <martinvonz at google.com>
>> >> # Date 1498281322 25200
>> >> #      Fri Jun 23 22:15:22 2017 -0700
>> >> # Node ID d654eefeefdd3df493fa777cbcfe423f17d9d500
>> >> # Parent  fac9941bd542e3621082b157fb2f3aff09cfb1b7
>> >> dagop: raise ProgrammingError if stopdepth<=0
>> >>
>> >> revset.py should never send such a value.
>> >>
>> >> diff --git a/mercurial/dagop.py b/mercurial/dagop.py
>> >> --- a/mercurial/dagop.py
>> >> +++ b/mercurial/dagop.py
>> >> @@ -33,7 +33,7 @@
>> >>      if stopdepth is None:
>> >>          stopdepth = _maxlogdepth
>> >>      if stopdepth <= 0:
>> >> -        return
>> >> +        raise error.ProgrammingError('negative stopdepth')
>> >
>> > I think stopdepth = 0 is valid on API level, so I want to change the condition
>> > to stopdepth < 0. I'll update it in flight if you agree.
>>
>> Weren't you passing in "stopdepth=(n + 1)" where n was a non-negative
>> number? I.e. n must be >= 0, so stopdepth > 0.
>
> Yes. There's no caller that passes stopdepth=0. I mean
> revancestors(..., stopdepth=0) => {} isn't an invalid use of the API. Oh, I see what you're saying. It would just no yield any matches, right? So it would be:

if stopdepth == 0:
  return
if stopdepth < 0:
  raise ProgrammingError()

Did I get that right?