Hi, On Sun, Nov 20, 2005 at 05:45:13PM +0100, Marcus Meissner wrote: > I have this patch proposal, but I fear it might not fly with Alexandre ;) Not only with Alexandre, I'm afraid ;)
Why not follow a dual strategy: use a static buffer for <= MAX_PATH and use a malloc()ed buffer if it exceeds that size? (ok, scrap that, static plus malloc() is a BAD idea:) Or, since the original buffer variable was static anyway, you could just *always* (re-)use an malloc()ed buffer and *grow* it whenever the previous size was to small... I haven't looked too closely at the code, but repeated malloc()/free() in "fast-as-lightning" registry code sounds like a terrible solution that can be easily transformed into something better here, it seems. Andreas
