Re: [PATCH] Lose perl dependency. (fwd)

From: Junio C Hamano <junkio@cox.net>
Date: Jan 20, 2007, 00:35 UTC
Message-ID: <7vvej2bkn2.fsf@assigned-by-dhcp.cox.net>
In-Reply-To: <Pine.LNX.4.63.0701200052210.12889@wbgn013.biozentrum.uni-wuerzburg.de>

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

Show 13 quoted lines

>> I think this is sane but I hate to having to worry about
>> possible fallouts from giving --reverse in setup_revisions() to
>> make it available to everybody.  E.g. things like "what happens
>> when you say "git format-patch --reverse HEAD~3".
>
> It would
>
> 	1) traverse all commits, storing them in a commit_list,
> 	2) reverse the commits, and then
> 	3) continue as before.
>
> So I don't really see a problem (after all, you don't have to use it if 
> you don't want to).

Well, I understand what the code does, but what does the above three steps MEAN to the end users?

In other words, if it does not make sense for format-patch to take --reverse, maybe we should keep it as an internal option, just like git-show is the only user of no-walk. And give option parsing for it for only selected commands (like rev-list) where it makes sense.

I am sure you can come up with a reason why the above three steps are useful for the end user, and it could turn out to be a very valid reason.

But format-patch was just one example. I will have to worry about all the users of revision traversal machinery.

The end result might be "ok, we have spent quite a lot of time and audited every users of revision machinery and for all of them --reverse has some valid use cases." and that would be wonderful. But the thing is, I hate to having to worry about that right now.