Re: Proposal for write(1) addition

[christos%astron.com@localhost (Christos Zoulas)]

In article <201109172253.SAA02263%Sparkle.Rodents-Montreal.ORG@localhost>,
Mouse  <mouse%Rodents-Montreal.ORG@localhost> wrote:
>>> Um, note the wording: a _utmp-writing_ daemon.  This would be for
>>> creating (and presumably cleaning up) utmp entries, not for
>>> backending write(1).
>> We don't want one or need a daemon; just use pututxline().
>
>That's a bit like saying, we don't need a kernel, just use open().
>This would be one possible implementation of pututxline() - a nice one,
>in some respects.  (Less nice in others, of course.  This is not to say
>that the existing NetBSD implementation of pututxline() is not also
>nice in some respects.)

pututxline() runs a setuid program to update utmp if needed, so it is
transparent to the user.

>> We also don't need setuid to deal with ptys.  This is what ptyfs is
>> for.
>
>This sounds as though you just decreed that ptyfs is required to use
>ptys (at least, by implication, for non-root, and, by context, if you
>care about utmp entries).  Correct?

No, ptyfs has nothing to do with utmp, and yes, if you want to use ptys
properly without requiring root, you need ptyfs. You also avoid the scan
the ptys code and the exposure of the control node. I don't see any advantages
using the old BSD ptys, do you?

christos