Re: [RFC/PATCH] Change "refs/" references to symbolic constants

From: Andy Parkins <andyparkins@gmail.com>
Date: Dec 17, 2006, 09:00 UTC
Message-ID: <200612170900.19856.andyparkins@gmail.com>
In-Reply-To: <7vslffcy80.fsf@assigned-by-dhcp.cox.net>

On Saturday 2006, December 16 21:44, Junio C Hamano wrote:

> This is not rejected nor forgotten but has not been applied yet,
> and may not be immediately.

Oh good.

> I would love to have this; it would reduce mistakes (spelling
> "refs/head/" would silently compile fine and result in a broken
> command, while spelling PATH_REFS_HEAD would be caught by the
> compiler).

I'd not even considered that; I liked it because of a natural disinclination towards repeated use semanitcally identical literals.

> A sad thing is this needs to be done when things are relatively
> quiescent.

While true - is there ever such a time?  :-)

On the other hand, this is (hopefully) a zero-impact change, and should leave the object the same... I might even test it by comparing objects.

> So I would apply this before v1.5.0 (or a replacement, if you
> have updates), but it might have to wait a bit until I I can say
> "ok, no big patches to C part is pending right now and it is a
> good time to do this clean-up".

If I know it's wanted I will happily maintain it indefinitely. It's part of my rebase on master set anyway, so it's no pain to do so. I'll keep an eye out for quiet times, or if you prod me when you want it I'll make you a rebased version instantly.

While we're on this subject, does the same apply to my hashsize literals patch which swapped 20 for HASH_WIDTH?

Andy

-- 
Dr Andrew Parkins, M Eng (Hons), AMIEE