Bug 510711

Unless I'm mistaken, we should no longer need python-2.*-canonicalize.patch that's being forward-ported since FC4 days.  It was added to address an issue reported in bug #169046, but never made it upstream.  The patch seems littler over-engineered and should no longer be needed, as I explain in:

  https://bugzilla.redhat.com/show_bug.cgi?id=169046#c8

I'd appreciate if someone can give it a second look and correct me if I'm wrong.


+++ This bug was initially created as a clone of Bug #169046 +++

I know this is ancient, but can anyone still remember further details about this?  I'd really love to understand why such a complex patch was created to address this, when one-liner as:

--- Python/sysmodule.c.canonicalize
+++ Python/sysmodule.c.fixed
@@ -1178,7 +1178,7 @@ void
 PySys_SetArgv(int argc, char **argv)
 {
 #if defined(HAVE_REALPATH)
-	char fullpath[MAXPATHLEN];
+	char fullpath[PATH_MAX];
 #elif defined(MS_WINDOWS)
 	char fullpath[MAX_PATH];
 #endif

should be good enough?  The problem here was that python was calling realpath with buffer of insufficient size.  The reason for that was that python defines MAXPATHLEN and does not use one defined on Linux in sys/param.h.  Python used to defined it unconditionally to 1024, regardless of the PATH_MAX value, which is the size expected by realpath.

The canonicalize.patch we use never made it upstream, but other changes resolved the issue there.  First, following change was done:

  http://svn.python.org/view?view=rev&revision=45715

MAXPATHLEN is no longer unconditionally set to 1024, but rather to PATH_MAX, if it's defined and is more than 1024.

Additionally, in python3, PATH_MAX is used as size of the buffer passed to realpath (see _wrealpath() in Python/sysmodule.c).