Re: hurd/term users.c

[Thomas Bushnell, BSG]

Roland McGrath <roland@frob.com> writes:

> Why do you say that?  I have never known of such behavior.  Linux does not
> do it.  FreeBSD does not do it.  Those are the systems that it's handy for
> me to check right now, but I don't know of any reason to ever have
> suspected such a behavior.  There is nowadays some standard that specifies
> select, though I only have 1003.1-1996 (which does not).  I would be
> surprised if it required, or even allowed, such an error diagnosis
> given the state of extant systems.

Well, the Linux manpage on my Debian system for select has:

       EBADF  An  invalid file descriptor was given in one of the
              sets.

I interpreted that to include both a totally bogus number as well as
one being used inappropriately.  But I guess it's ok to let them
through silently.

Linux pipe code, for example, will let you have a pipe open only for
writing, and still poll on read for that pipe, AFAICT.  That's so
totally wrong; we should not duplicate that.  

So I don't mind replacing such errors with "no response" results; but
I strongly object to letting people getting poll results for read (or
write) on an fd that they don't have open for reading (or writing).
Replacing the error case with a "just return no" case is fine by me,
but don't just elide the check entirely.