From: Junio C Hamano <junkio@cox.net>
Date: Wed, 06 Sep 2006 04:46:50 GMT
Subject: Re: [PATCH 1/2] rev list add option accepting revision constraints on standard input
Message-ID: <7vk64ha9fp.fsf@assigned-by-dhcp.cox.net>
In-Reply-To: <7vpseaarrl.fsf@assigned-by-dhcp.cox.net>

Junio C Hamano <junkio@cox.net> writes:

> Is calling setup_revisions() on the same revs like this many
> times safe?  I do not think so, especially what is after the
> primary "for()" loop in the function.
>
> I was sort-of expecting that you would instead replace that
> primary for() loop in setup_revisions() with some sort of
> callback...

I take half of the above back.  Even after setup_revisions()
returns, adding more revisions via add_pending_object() is
safe.  However, the postprocessing done in setup_revisions()
after its main loop, while I do not think they would crash when
called twice, would be very wasteful.

And ``callback'' interface is usually very cumbersome to use, so
we probably would want to avoid it unless absolutely necessary.

I've outlined an alternative implementation in two patches;
I'll be sending them out shortly.