[2206] in Kerberos_V5_Development
Re: Comerr library problems....
daemon@ATHENA.MIT.EDU (Theodore Y. Ts'o)
Thu Feb 6 14:18:51 1997
Date: Thu, 6 Feb 1997 14:11:42 -0500
From: "Theodore Y. Ts'o" <tytso@MIT.EDU>
To: epeisach@MIT.EDU
Cc: basch@lehman.com, krbdev@MIT.EDU
In-Reply-To: epeisach@MIT.EDU's message of Thu, 6 Feb 1997 11:44:22 -0500,
<9702061644.AA01328@kangaroo.mit.edu>
From: epeisach@MIT.EDU
Date: Thu, 6 Feb 1997 11:44:22 -0500
Cc: krbdev@MIT.EDU
In trying to build the current source tree, the com_err library has
changed. com_err_va() used to exist - and is documented in
com_err.texinfo. It is used in util/ss/error.c...
Ah, oops. I'll deal with that. com_err_va wasn't defined in com_err.h,
so I assumed it was an internal function only.
There appears to be an et_ctx structure now. As this is not documented,
would the proper "fix" be to call et_com_err_va with the first two
arguments of 0?
Note that the et_ctx structure is an experiment only. The "V2 API" was
something that I did more as a proof of concept than anything else, and
shouldn't be used at this point.
Or maybe adding a compatibility function in util/et/com_err.c
com_err_va() which DTRT...
Adding a compatibility function is the right thing; I'll be doing that.
The new com_err library is supposed to have all of the original API's
supported by the old com_err library. The main thing that I've done is
started cleaning it up, and put in the et_ctx experiment (which may
disappear later).
Also, in com_err.c, com_err() or et_com_err() is defined depending on
whether the machine has stdarg.h - is this a typo? (look for ET_VARARGS).
That's a typo; thanks for catching it....
- Ted
