Mail Archives: cygwin/2008/10/31/09:01:41
On Fri, Oct 31, 2008 at 9:04 AM, Eric Blake wrote:
> POSIX 200x (which is on track to be approved later this year) has modified
> the requirements on realpath() that NULL be a universally accepted
> argument.
Oh, good.
> Not all vendors comply with this yet, but at some point, it
> will no longer be a GNU extension, but a POSIX-mandated requirement that
> realpath() be able to malloc its result. This is part of the overall
> paradigm change in POSIX 200x that recognizes that it is legal to have
> paths longer than PATH_MAX
I haven't tested it, but IIRC, GNU's version doesn't currently
support such paths, either. It allocates the buffer for you if you
pass in NULL, but it still only allocates PATH_MAX bytes.
Thanks for the info. Anything to improve that interaction is a
welcome change. The idea of a "maximum path length" is demonstrably
bogus, while string functions that take a destination buffer but no
size parameter are just plain dangerous.
--
Mark J. Reed <markjreed AT gmail DOT com>
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
- Raw text -