[6] in Kerberos_V5_Development

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

comments on header files

jtkohl@ATHENA.MIT.EDU (jtkohl@ATHENA.MIT.EDU)
Mon Dec 4 15:09:15 1989

Date: Mon, 27 Nov 89 11:42:56 -0500
From: John T Kohl <jtkohl@ATHENA.MIT.EDU>
To: sommerfeld@APOLLO.COM
In-Reply-To: Bill Sommerfeld's message of Wed, 22 Nov 89 14:41:15 EST,
	<8911221948.AA07674@xuucp.ch.apollo.com>
Subject: Re: comments on header files.

> Date: Wed, 22 Nov 89 14:41:15 EST
> From: sommerfeld@APOLLO.COM (Bill Sommerfeld)

> <mit-copyright.h> is 15 characters long.  I'd like to be able to deal
> with brain-damaged operating systems more cleanly (mumble mumble HP-UX
> grumble grumble AIX barf bletch).  Since the compiler doesn't care

I've changed it to <copyright.h>; I prefer to leave it doing at least
that until I get a reading/opinion from lawyer types.

> Can you make the file names of the .h files and the #ifndef/#define
> .. #endif tags match better in all cases (base-defs.h doesn't match
> very well).

done; all are related to filenames now, and all are __KRB5_xxx__

> Re: encryption.h:

> The encryption/decryption interface seems to assume a block cipher,
> not a (blocked-)stream cipher; I think the latter might be a better
> choice of interface.

The problem is that assumption of a stream cipher causes some problems;
assuming a block cipher is a more general case of a stream cipher (which
can be considered a block cipher with blocksize 1).

> "krb5_keyblock" is variable length, but appears in-line in
> krb5_encrypt_block.

changed to a pointer out of krb5_encrypt_block.

> why not just pass a pointer to the encrypt_block?

Done.

John

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