Re: [PATCH] Revert "graph.c: mark private file-scope symbols as static"

From: Junio C Hamano <gitster@pobox.com>
Date: Mar 3, 2013, 22:49 UTC
Message-ID: <7vppzg9k0n.fsf@alter.siamese.dyndns.org>
In-Reply-To: <20130303214206.GL7738@serenity.lan>

John Keeping <john@keeping.me.uk> writes:

Show 17 quoted lines

> On Sun, Mar 03, 2013 at 01:08:50PM -0800, Junio C Hamano wrote:
>> >> > Additionally, it seems that Johan added graph_set_column_colors
>> >> > specifically so that CGit should use it - there's no value to having
>> >> > that as a method just for its use in graph.c and he was the author of
>> >> > CGit commit 268b34a (ui-log: Colorize commit graph, 2010-11-15).
>> >> 
>> >> Perhaps you could add a comment in the source to prevent this from
>> >> happening again?
>> > ...
>> > I would hope that having this message in the history should be enough to
>> > prevent this changing in the future....
>> 
>> Given how it happened in the first place, I do not think anything
>> short of in-code comment would have helped.  There wouldn't be any
>> hint to look into the history without one.
>
> So you'd accept a patch doing that?

The answer obviously depends on the specifics of "that" ;-) I was merely agreeing with what Thomas said. A straight-revert would be insufficient to prevent this from recurring again.

> Something like this perhaps:
>
>     NOTE: Although these functions aren't used in Git outside graph.c,
>     they are used by CGit.

It would be a good place to start, although I prefer to see it completed with s/used by CGit/& in order to do such and such/ by somebody working on CGit.

Also it probably is worth adding contact information for folks who work on CGit (http://hjemli.net/git/cgit/ might be sufficient), as changing these functions (e.g. changing the function signature) will affect them; making them "static" is not the only way to hurt them.

Thanks.