Re: [PATCH 1/2] terminal: teach git how to save/restore its terminal settings

From: Junio C Hamano <gitster@pobox.com>
To: Carlo Arenas <carenas@gmail.com>
Cc: git@vger.kernel.org, johannes.schindelin@gmx.de
Subject: Re: [PATCH 1/2] terminal: teach git how to save/restore its terminal settings
Date: Mon, 04 Oct 2021 11:10:49 -0700	[thread overview]
Message-ID: <xmqqlf38wthy.fsf@gitster.g> (raw)
In-Reply-To: <CAPUEsph_Aqaveur4qZ+Z+85gw1z2CvPOkzbkqW9zt56RoHhUJw@mail.gmail.com> (Carlo Arenas's message of "Mon, 4 Oct 2021 10:27:06 -0700")

Carlo Arenas <carenas@gmail.com> writes:

> On Mon, Oct 4, 2021 at 9:36 AM Junio C Hamano <gitster@pobox.com> wrote:
>> Carlo Marcelo Arenas Belón  <carenas@gmail.com> writes:
>> > diff --git a/compat/terminal.c b/compat/terminal.c
>> > index 43b73ddc75..1fadbfd6b6 100644
>> > --- a/compat/terminal.c
>> > +++ b/compat/terminal.c
>> > @@ -8,7 +8,7 @@
>> >
>> >  #if defined(HAVE_DEV_TTY) || defined(GIT_WINDOWS_NATIVE)
>> >
>> > -static void restore_term(void);
>> > +void restore_term(void);
>>
>> Curious why you need this because (1) we do not have the same for
>> save_term() here, and (2) we include compat/terminal.h where these
>> two are declared next to each other.
>
> It is in preparation for the next patch where we will call these newly
> public functions from editor.c
> I'll be reusing restore_term(), while save_term() is new, hence why
> only one had this change.

I think I understand all that correctly.

I was curious why the patch left a forward declaration, instead of
just removing it, which it can do because now we have the necessary
declaration in the header file it includes.

With only [1/2]:

 - we already have save_term() and restore_term() externally
   declared, so even outside users can use them (which is a good
   thing to do), as long as they include <compat/terminal.h>.

 - we include <compat/terminal.h> in compat/terminal.c; I do not see
   why the patch needs to turn a static forward declaration into an
   extern one in the hunk in question.

Thanks.