Re: UID defined on an NIS server.

[Owen Taylor <otaylor redhat com>]

"Ian King" <king eiffel com> writes:

> Actually this is not my system but a relay of bug reports from various
> people using linux EiffelVision 2.  Under what circumstances should this
> fail?  I presumed that both bug submitters were using NIS but actually it is
> only one that seems to be,  how could this possibly fail anyway?  If a user
> manages to login and is assigned a UID, then surely getpwuid_r will never
> fail?

Hmmm, I'm remembering some problems with systems having broken getpwuid_r;
though the GLib configure.in checks are pretty exhaustive - they deal
with the case where there is a function, but it returns ENOSYS.

Of course, without details about the systems where this occurred, it's
pretty hard to say.

A bug submitted bugzilla.gnome.org might be a good way to track this in
issue in any case... though it's pretty clearly a system problem, not
a GLib problem, we are already doing a lot of gymnastics to try and
make getpwuid_r work portably and catch brokeness.

Regards,
                                        Owen