"Jeremy C. Reed" <reed@reedmedia.net> wrote:

> As I mentioned in my email: software probably shouldn't use the actual
> numerical UIDs or GIDs, but should use getpwnam(3) or getgrnam(3) or
> similar.

Same problem, just the other way around.  What do you do if you have a
user called, say, "qmail" (Quentin J. Maileron, say UID 4387) on one of
the systems?  I was always under the impression that the whole concept of
UID vs. symbolic name makes it a requirement for programs to *never*
rely on the user_name_, but rather the numerical UID.

I wouldn't feel quite right about programs that check for the username
"root" rather than UID 0.

Am I dense?  Do I miss something here?

-Jan

-- 
finger jschauma@netmeister.org