From: Junio C Hamano <gitster@pobox.com>
Date: Wed, 11 Mar 2026 19:35:26 GMT
Subject: Re: [PATCH 0/4] wean start_command() off the_repository
Message-ID: <xmqqjyvinoqp.fsf@gitster.g>
In-Reply-To: <041604ef-0bdb-4c40-a9e1-60c38ae664cb@web.de>

René Scharfe <l.s.r@web.de> writes:

> On 3/11/26 4:19 PM, Burak Kaan Karaçay wrote:
>> 
>> start_command() relies on the_repository due to the 'close_object_store'
>> flag in 'struct child_process'. Introduce repo_start_command() to allow
>> working with arbitrary repositories. Turn start_command() into a macro
>> that wraps repo_start_command() and migrate the existing callers with a
>> cocci script.
>
> Good idea to expose this hidden dependency.  It's different from the
> other repo_* functions, though, in that most callers can safely pass
> NULL as repo because they don't set close_object_store.
>
> Only gc, pull and auto-maintenance set close_object_store.  If we
> changed them to set a pointer to the object store they want to have
> closed instead of a binary flag then we could leave the other callers
> unchanged.

You solved my "don't we need to know where we are coming from, in
addition to where we are going?" question elegantly.

Thanks.