[PATCH] hgext/mq - idempotent operations should return success

[Peter Williams]

Peter Williams wrote:
> Alexis S. L. Carvalho wrote:
>> Thus spake Peter Williams:
>>> Alexis S. L. Carvalho wrote:
>>>> I mostly agree with this, but I really think that qpush without
>>>> arguments should return an error if it can't push another patch.
>>>> Otherwise you'll break this usage:
>>>>
>>>>    while hg qpush; do
>>>>        something
>>>>    done
>>>>
>>>> Same thing for qpop - but the error message in both cases could 
>>>> probably
>>>> be more friendly (see issue474).
>>>>
>>>> Patches welcome :)  (especially if you can use hg export to generate
>>>> them with author and commit message).
>>> Do you have some mq patches applied to your repository?  The reason 
>>> that I ask is that when I try to clone it get:
>>
>> Hmm?  Depends on the repository
> 
> The on advertised on your web site.
> 
>> - I don't really a have a canonical one,
>> and I usually only make them available on HTTP when I want to make some
>> code available for review...  Patches on top of either the main[1] or
>> the crew[2] repo are usually fine...
>>
>>> abort: prechangegroup.mq-no-pull hook exited with status 1
>>
>> The prechangegroup hook is called by the local hg before adding the
>> revisions to the repo.  IOW, it looks like a problem on your end.
> 
> I must admit that I did find it confusing but I'm doing a clone (not a 
> pull) into a directory that isn't a hg repository and assumed that clone 
> might be testing the source repository in some way.
> 
>>
>> If this is the same "prechangegroup.mq-no-pull" from the wiki, it should
>> only fail if the repo you're pulling into has mq patches applied, which
>> is never the case for clone...
> 
> See above re confusion.
> 
>>  Was this with an unmodified hg?
> 
> Yes.
> 
>>  This
>> hook depends on the exit status of hg qtop...
> 
> I've tried the clone from a different machine and it works OK so it must 
> be a problem with my set up on the first machine.  I'll look into it.
> 
> Thanks for your help
> Peter

Problem solved.  The machine with the problem was running Fedora-5 and 
hence had mercurial-0.9.1 installed as that is the latest available for 
installation for Fedora-5.  I manually built and installed a 0.9.3 rpm 
and everything works OK.

Perhaps Fedora-5 needs to be updated to 0.9.3.  I also noticed that 
freshmeat lists 0.9.2 as the latest release of mercurial.

Peter
-- 
Peter Williams                                   pwil3058 at bigpond.net.au

"Learning, n. The kind of ignorance distinguishing the studious."
  -- Ambrose Bierce