Re: [PATCH] builtin/clone: avoid failure with GIT_DEFAULT_HASH

From: Junio C Hamano <gitster@pobox.com>
Date: Sep 14, 2020, 21:44 UTC
Message-ID: <xmqqk0ww119u.fsf@gitster.c.googlers.com>
In-Reply-To: <xmqqo8m811mf.fsf@gitster.c.googlers.com>

Junio C Hamano <gitster@pobox.com> writes:

Show 16 quoted lines

> "brian m. carlson" <sandals@crustytoothpaste.net> writes:
>
>>> Alternatively, I certainly wouldn't complain if you did introduce a new
>>> function and updated the call-site that passes reinit as 1.
>>
>> I thought about introducing a new function, but since it would share
>> almost all of the code, it seemed a bit wasteful, even if the function
>> is small.  We do have only two callers, I believe, since I recall
>> making this function non-static and calling it from clone, so I think
>> it's okay.
>
> Perhaps.
>
> FWIW, this seems to have strange interaction with something in
> 'seen' when merged; I suspect it is the topic that mucks with the
> set-up sequence for "git clone", but didn't check.

Actually I have to take it back. I have this directly on top of v2.28.0 and it already breaks tests big time. For example, here is how "cd t && sh t0021-conversion.sh -i -v" breaks:

    ...
    Cloning into 'repo-cloned'...
    fatal: could not unset 'extensions.objectformat'
    fatal: the remote end hung up unexpectedly
    not ok 25 - delayed checkout in process filter

The story is the same if it is applied on top of 'master' (which is expected, as we haven't done anything to affect this area since v2.28.0).