On Thu, Nov 13, 2008 at 5:31 AM, Julian Foad <julianfoad_at_btopenworld.com> wrote:
> On Thu, 2008-11-13 at 04:47 -0800, Greg Stein wrote:
>> Why not just change the function signature and the callers?
>
> The thought crossed my mind, but a code change takes more care and
> testing, and we're trying to stabilise and concentrate on bugs. That's
> why I didn't do it.
>
> I take your point about being consistent with the way "loggy" functions
> take a pointer-to-pointer and allocate if necessary. You pointed this
> out to me before about some semi-public APIs and I agreed and changed
> them. However, in this file it is common for local functions to take a
> pointer-to-existing-buffer, and the callers all have a log accumulator
> already (one of them appears to create it specifically for this
> purpose).
Fair enough. Had to ask :-)
As I may have mentioned before: I think it would have been nicer for
the functions to simply take a pointer, and expect the caller to
create the buffer. The caller has to do *something* (init to NULL or
alloc), so it may as well go with the simpler signature.
But... right now, we're stuck with this double-indirect form for the
bulk of the code. It will get fully revamped in 1.7 when we move the
loggy operations into the database.
Cheers,
-g
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-11-13 20:29:01 CET