On Jun 19, Michael Banck ([EMAIL PROTECTED]) wrote: > Hi, > > On Sun, Dec 25, 2005 at 08:05:40PM -0500, Neil Roeth wrote: > > I'll upload what I have so that the new release can get out there. > > Then, I'll work with you after you get back from vacation to fix the > > hurd bug, which I'd like to do using realpath() with a NULL second > > arg. This is apparently the best way to do it on an architecture that > > has no limit on the path length. > > So, we took another look at this, and came up with this patch which is > hopefully the least intrusive. > > It makes one assumption, though, namely that systems without MAXPATHLEN > will have a realpath() supporting NULL as second argument. Or at least, > it only fixes it for those systems. I know of no systems which neither > have MAXPATHLEN nor are using glibc. If you think this is not totally > kosher, we could add some #ifdef(__GLIBC__) maybe, but I don't think > this is necessary and would only make the code less readable. > > Another thing is that realDirs and realOutputDir only get free'd when > the checks succeed. This should not matter as otherwise exit(1) gets > called, but might make valgrind unhappy. If this is unacceptable for > opensp, I can update the patch to also free() them in that case.
Thanks, I'll put in this patch. -- Neil Roeth -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
