Re: [PATCH v5 08/15] remote-testgit: cleanup tests

[Felipe Contreras <felipe.contreras@gmail.com>]

On Wed, Nov 21, 2012 at 7:28 PM, Junio C Hamano <gitster@pobox.com> wrote:
> Felipe Contreras <felipe.contreras@gmail.com> writes:
>
>> We don't need a bare 'server' and an intermediary 'public'. The repos
>> can talk to each other directly; that's what we want to exercise.
>
> The previous patch to remove the test (the one that covered a case
> where a bug was fixed in an older git-remote-testpy and tried to
> catch the bug when it resurfaced) made sense even with its
> ultra-short justification "irrelevant".
>
> But I am not sure if this one is so cut-and-dried.  The repos can
> talk to each other directly, but at the same time the tests were
> exercising interactions between bare and non-bare repositories,
> weren't they?  Talking to each other may be one of the things we
> want to exercise, but that does not necessarily be the only thing.
>
> If it were explained like this (note that I am *guessing* what you
> meant to achieve by this patch, which may be wrong, in which case
> the log message needs further clarification):
>
>         Going through an intermediary 'public' may have exercised
>         interactions among combinations of bare and non-bare
>         repositories a bit more, but that is not an issue specific
>         to the remote-helper transfer that we want to be testing in
>         this script.  Simplify the tests to let two repositories
>         talk directly with each other.

Right. I don't think bare vs. non-bare has anything to do with it; the
intermediary repository was there to have 3 types of repos interacting
with each other local testpy, remote testpy, local git. But this
doesn't exercise anything from transport helper.

Cheers.

-- 
Felipe Contreras