[16271] in Athena Bugs

home help back first fref pref prev next nref lref last post

Re: olc_answers

daemon@ATHENA.MIT.EDU (Albert Dvornik)
Sun Aug 30 23:26:48 1998

To: Camilla R Fox <cfox@MIT.EDU>
Cc: bug-olc@MIT.EDU, olc_adm@MIT.EDU
From: Albert Dvornik <bert@MIT.EDU>
In-Reply-To: Camilla R Fox's message of "Sat, 29 Aug 1998 00:18:11 EDT"
Date: Sun, 30 Aug 1998 23:26:29 EDT


Camilla R Fox <cfox@MIT.EDU> writes:
> On the dialups (and apparently other 8.2 solaris machines) olc_answers
> seems to have problems.
[...]
> #!/bin/sh
> exec /bin/athena/attachandrun olc-stock olc_browser "$0" "$@"

Oy.  Apparently invocation for a bunch of attach-and-run scripts on
the packs was changed for 8.2, and no one bothered to test all of
them.

> olc_browser (either location) runs fine, and also runs fine if you give
> it random arguments.  The problem appears to show up if I give it
> anything other than olc_browser as $0, with different breakage showing
> if I put olc there.
[rest of hypothesis deleted]

It has to do with the fact that `olc_browser', `owl_browser' and
`olta-browser' are all the same binary, and giving it `olc_answers'
confuses the selection mechanism.

> Failing that, it would be trivial to modify the olc_answers script to
> make $0 olc_browser.

*** bugs:
Yes, I think that's the way to go.  [Specifically, if the third field
of packs/glue/specs is specified, I believe that rather than $0 should
be used for the argv0.  Greg, if you agree, a patch is below.]

Also, let me know if you think oXX_browser should accept oXX_answers
as its argv[0], and we can have a civilized argument about it. =)

*** olc_adm:
Until this gets patched in the release, I think a temporary fix in the
olc-stock locker is in order.  I'd have done it, but I didn't have the
bits in the appropriate binary directories...

> (I failed to find the source, or any documentation for the browser
> itself, after a half hearted search, so otherwise, I'm stumped.)

FYI, the source to olc_browser is in the OLC sources.  The source to
the current olc_answers is generated from /mit/source/packs/glue.
Thanks for reporting this!

--bert

home help back first fref pref prev next nref lref last post