Re: [bug-inetutils] dnsdomainname, domainname, nisdomainname, ypdomainname

[Mats Erik Andersson]

måndag den 12 mars 2012 klockan 22:01 skrev Simon Josefsson detta:
> address@hidden (Alfred M. Szmidt) writes:
> 
> >    > Looks good; but after looking at the actual name for
> >    > dnsdomainname several times now I cringe...  Domain Name Service
> >    > Domain Name?!  Does domainname do the same as dnsdomainname? Can
> >    > we use that instead?
> >
> >    The semantics of these four tools are more or less set in stone,
> >    for compatibility with existing tools.  The terminology in this
> >    area is indeed confused but given the background I can't see what
> >    we can do more than making sure the documentation is complete.
> >    Simple memory rule:
> >
> > I was afraid of that, oh well... Anyway, go go go Simon go!
> 
> I have pushed 'dnsdomainname' now.  Let's take a look at the others
> next...

The test script is problematic:

  a) There is no requirement that the name of any single computer
     is set to something that is resolvable by DNS or any other
     method. If that information is lacking, then "dnsdomainname"
     fails with "cannot determine host name", and the test script
     gives a false negative. This happens for me with NetBSD being
     used with Qemu.

  b) A particular system may be legally set up in a manner that
     a non-fully qualified host name is returned by xgethostname()
     and ensuing resolution. This can be accomplished by

        # /etc/hosts
        10.0.1.1    dag  dag.utan.ljus

        # /etc/nsswitch.conf
        hosts   files dns

        $ hostname
        dag

     Thus "dnsdomainname" fails with "cannot qualify host name: dag",
     again producing a false negative in the test script. This happened
     to be the case on my OpenSolaris/OpenIndiana systems!

The test script should report "skipping" in either of the above cases.

Regards,
  Mats