[1513] in Kerberos-V5-bugs
Re: autoconf
daemon@ATHENA.MIT.EDU (epeisach@MIT.EDU)
Wed Jul 5 13:04:25 1995
From: epeisach@MIT.EDU
Date: Wed, 5 Jul 1995 13:04:15 -0400
To: ejb@era.com
Cc: krb5-bugs@MIT.EDU
I cannot speak for the policy regarding responses to bug reports, etc. I
know that when someone reports a problem and then it is fixed in the
source tree they have been getting messages. Unfortunatly, many of the
older reports no-longer apply due to autoconfing.
Regarding your comments on autoconf. In the next release, autoconf.h
should not be needed, except in the build tree. If you, as a developer,
include krb5.h, everything you need should be there - error table
definitions, etc. The k5-int.h, ext-proto.h, etc, probably are not
really useful unless you are in the krb5 tree. What is being done, is
that operating specific definitions which are required for use of krb5.h
are being merged in. (i.e. SIZEOF_LONG, etc.) This in essence requires
all the subsequeunt Makefiles (especially in appl) to -DFOO all the
particulars for the application. Shipping a krb5-config.h with operating
system definitions (such as HAS_CRYPT) may provide a setup to lose if
developers come to rely on these standard autoconf definitions being
there when they may be removed later.
Regarding the telnet message: I saw a change made recently that might
have been related to that problem - I am not positive.
Regarding typos - I will try to get those in.
Ezra
