[5431] in Kerberos
Re: Question for the Kerberos Implementors
daemon@ATHENA.MIT.EDU (Marc Horowitz)
Tue Jun 27 17:50:52 1995
To: kerberos@MIT.EDU
Date: 27 Jun 1995 21:35:04 GMT
From: marc@cam.ov.com (Marc Horowitz)
In article <199506211550.IAA12883@elrond.Eng.Sun.COM> Danny.Nessett@Eng.Sun.COM (Dan Nessett) writes:
Here gss_release_oid_set() is called instead of the "internal" entry point
generic_gss_release_oid_set(). This is causing the object file krb5_gss_glue.o
to be called in by the linker, which holds all of the gss_ routine entry
points. As a result, I am getting multiple definition errors during the
link stage. This problem can be easily fixed by changing the reference to
generic_gss_release_oid_set().
Could I have a reading from the implementors on the layering objectives for
the Kerberos V5 code? If this is a bug, I will file it at
krb5-bugs@mit.edu
Thanks,
Dan
The intent is as you surmise. I wanted to make it possible to use the
mechanism in a multimech implementation, which is why I have the
krb5_gss_glue.c file the way it is. It looks like you found a
mistake. If that's the only one, then I did better than I thought,
since I never actually tried to use the code this way :-)
You should file a bug.
Marc
