Re: [PATCH v4] Expand ~ and ~user in core.excludesfile, commit.template

From: Junio C Hamano <gitster@pobox.com>
Date: Aug 29, 2008, 19:28 UTC
Message-ID: <7vk5dz4o3t.fsf@gitster.siamese.dyndns.org>
In-Reply-To: <quack.20080829T1201.lthsksnir1u@roar.cs.berkeley.edu>

Karl Chen <quarl@cs.berkeley.edu> writes:

Show 11 quoted lines

>>>>>> On 2008-08-29 09:08 PDT, Junio C Hamano writes:
>
>     Junio> I do not see any strong reason why the single caller of
>     Junio> user_path() has to keep using the static allocation.
>     Junio> Would it help to reduce the complexity of your
>     Junio> expand_user_path() implementation, if we fixed the
>     Junio> caller along the lines of this patch (untested, but
>     Junio> just to illustrate the point)?
>
> Yes, expand_user_path() would be much simpler, it would basically
> be me original implementation except for returning NULL on error.

Yeah, modulo those styles issues your v3 and v4 addressed, and use of strbuf.

It might feel that we went full circles, wasting your time. But it's not. We found out that the final series would look like this:

 [1/3] Introduce expand_user_path();

 [2/3] Using #1, introduce git_config_pathname() and use it to parse your
       two variables;

 [3/3] Update the sole caller of user_path() to use expand_user_path().

Patch #1 and #2 can be squashed into one if you want. Also you do not have to do #3 yourself if you do not feel like it (but now we know how the code would look like, why not?).

Thanks to these three initial rounds, we know whoever implements #1 knows what kind of interface the (to-be-rewritten) user of user_path() would want, so #3 will become much cleaner. We made progress.

Thanks.