From: Sibi Siddharthan <sibisiddharthan.github@gmail.com>
Date: Wed, 13 May 2020 19:57:50 GMT
Subject: Re: [PATCH v2 10/11] cmake: parse the makefile for the sources.
Message-ID: <CAKiG+9Uc8Kc1F3wN7AoLkBR8Uk1up5sZXpgUperhQO6NVmBRxw@mail.gmail.com>
In-Reply-To: <xmqqk11gc1d5.fsf@gitster.c.googlers.com>

On Wed, May 13, 2020 at 2:33 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> "Sibi Siddharthan via GitGitGadget" <gitgitgadget@gmail.com> writes:
>
> > From: Sibi Siddharthan <sibisiddharthan.github@gmail.com>
> >
> > The CMake script parses the Makefile for:
> > SCRIPT_SH
> > SCRIPT_PERL
> > TEST_BUILTINS_OBJS
> > LIB_OBJS
> > BUILTIN_OBJS
> > XDIFF_OBJS
> > VCSSVN_OBJS
> >
> > By doing this we avoid duplication of text between the Makefile and
> > the CMake script.
>
> Thanks, this makes perfect sense, but it probably should have been
> done from the very beginning, without copying the huge list manually
> in the step [01/11].
>
> How robust is the "parser"?  Do we need to leave an instruction to
> Makefile writers to conform to certain style to help CMake users as
> an additional comment in the Makefile, or is any reasonable textual
> drift still accepted?
>

The parser just uses a regex capture like this "SCRIPT_SH += (.*)".
If the current source/script definition in the Makefile is continued
things will be fine.

On a side note, why are we using SCRIPT_SH and SCRIPT_LIB separately?
Can't we combine them into one? If we can then I can also add it to
the list above.

Thank You,
Sibi Siddharthan