Re: [tkined] Install Scotty 2.1.10 etc. (Digital Unix 4.0D)

Alan J. Flavell (flavell@a5.ph.gla.ac.uk)
Mon, 21 Jun 1999 20:05:31 +0100 (BST)

On Mon, 21 Jun 1999, Juergen Schoenwaelder wrote:

> Alan> This seems to refer to a mib file that I had added manually (and
> Alan> maybe clumsily?) in an earlier version... (2.1.7)

> Its your local fun. Perhaps something loaded from a .tnmrc file

Oh yes, it's in ~/.scottyrc - thanks for the clue. I think I've
sorted that point out now.

I just noticed a message from an earlier test in the "make check"
series. It doesn't actually say that the test failed, but I presume
this is considered a fault?

==== dns-3.1 dns address option
==== Contents of test case:

dns address localhost

==== Test generated error:
domainname error

Here's the corresponding effect with nslookup:

> nslookup localhost
Server: localhost
Address: 127.0.0.1

*** localhost can't find localhost: Non-existent host/domain

Kind'a weird?

> Alan> ==== netdb-3.6 check "netdb services" command ==== Contents of
> Alan> test case:
...
> Alan> ==== Result was: 753 != 752

> It just indicates some interesting definitions in your
> /etc/services (e.g. duplicate names/aliases).

Ah. I get it: the yellow pages (NIS) services entries are
incompatible with the /etc/services definitions. Nobody had spotted
that before! Thanks!!

> Alan> ---- netdb-3.6 FAILED sh: 8866 Memory fault - core dumped ***
> Alan> Exit 139 Stop.

> Core dumps are serious errors.

If I look at the test script for netdb, I see that after test 6.5
comes a test which is also numbered 6.5 (should be 6.6, evidently)
which is the one provoking the error. I omitted to mention that
there's a long delay (minutes) before it reports the crash.

If I issue the command in a scotty shell, again there is a long delay
and it crashes:

% netdb ip range 134.169.34.0 255.255.255.0
Segmentation fault (core dumped)

I tried reinstalling tk/tcl from 8.0.3 instead of 8.0.4 - since 8.0.3
was the one specifically mentioned in your instructions - and then
building your package from scratch - but the symptoms in the tests
seem to be the same.

By the way, both of the times I tried to configure the package "from
cold" it seems to have decided to use the prefix "/usr", rather than
the prefix "/usr/local" which I thought was meant to be the default.
I had to override it manually with ./configure --prefix=/usr/local
to get it to go where I intended it to go. Is this perhaps a mistake?

> However, you have to dig into the core
> in order to find out what went wrong.

OK, that's understood. However, the actual tkined etc. software seems
to be working for me, in a practical sense, which is what I need; I'm
happy to repay the compliment by doing what I reasonably can to help
debug the apparent problems being reported by "make check", but my
skills in actual code debugging at this level are rather limited, I'm
afraid. If you want to pursue this, perhaps we should discuss it off
the mailing list; or else I can just write it down to experience and
look forward to using a different platform (i.e either Solaris or
Linux) next time around ;-)

best regards, and thanks for this useful tool and for your help.

--
!! This message is brought to you via the `tkined & scotty' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <tkined-request@ibr.cs.tu-bs.de>.
!! See http://wwwsnmp.cs.utwente.nl/~schoenw/scotty/ for more information.