[319] in Kerberos_V5_Development
Re: Extension to krb5_unparse_name
daemon@ATHENA.MIT.EDU (Theodore Ts'o)
Thu Oct 11 14:27:00 1990
Date: Thu, 11 Oct 90 13:13:36 -0400
From: Theodore Ts'o <tytso@ATHENA.MIT.EDU>
To: John T Kohl <jtkohl@ATHENA.MIT.EDU>
Cc: krbdev@ATHENA.MIT.EDU
In-Reply-To: John T Kohl's message of Thu, 11 Oct 90 12:55:41 -0400,
Reply-To: tytso@ATHENA.MIT.EDU
Date: Thu, 11 Oct 90 12:55:41 -0400
From: John T Kohl <jtkohl@ATHENA.MIT.EDU>
X-Us-Snail: MIT Room E40-300, 1 Amherst St., Cambridge, MA 02139 USA
Before you put this in, let me examine the reasons from the meetings of
last fall/winter to see why we agreed to not worry about this quoting.
kprop/kpropd (well, actually the dump_database function that will be in
kdb5_edit), needs this functionality in a big way, since we can't
guarantee that components won't have '/', '@', or '\0' in them.
Currently, krb5_unprase_name returns an error if you try to unparse a
principal name with '/' or '@' in it, and it just does the wrong thing
if a null appears in it. I consider this to be a bug.
Frankly, I'd rather just diallow theose characters, (especially '\0'!)
since it's a much cleaner solution and would make the code a lot cleaner
and compact. Unfortunately, Kerberos 5 has already been hit by ISO / X11
disease....
The reason why we punted on quoting back in January was because people
were complaining about how ugly backquoting made things. We agreed to
put it off until some further time, because we were wasting large
amounts of time flaming about it on the meetings. Given that no one has
come up with a better solution since then, I'm proposing that my
backquoting routines be placed into the library --- they just add
additional functionality, since you can still use them without taking
advantage of the quoting. However, if you want to be able to parse and
unparse arbitrary principals with wierd characters in it, as might
appear in the Kerberos database, you're going to have to use my
routines.
- Ted
