[1421] in Kerberos_V5_Development
Re: On direct db access...
daemon@ATHENA.MIT.EDU (Theodore Ts'o)
Tue Jul 23 12:08:47 1996
Date: Tue, 23 Jul 1996 12:08:36 -0400
From: "Theodore Ts'o" <tytso@MIT.EDU>
To: epeisach@MIT.EDU
Cc: krbcore@MIT.EDU
In-Reply-To: <9607231314.AA23530@kangaroo.mit.edu> (epeisach@MIT.EDU)
From: epeisach@MIT.EDU
Date: Tue, 23 Jul 1996 09:14:43 -0400
I have a number of proposals:
a) Scrap the ndbm interfaces and rewrite in db..(probably will not go
over well)
Actually, I've changed my position about this one over time --- as long
as we use the dbopen(3) interface, we can always add a ndbm back end to
dbopen easily enough. The abstraction you're describing is pretty much
dbopen(3)'s, anyway, so why don't we just use it?
If it turns out that Marc hasn't done a complete job, and berk_db still
has a tendency to make a hash (sorry) out of Kerberosd databases, we can
always just plug in a different back end.
There is a question of _when_ we want to rewrite to db --- given that in
order to avoid screwing over Cygnus, we need to release Beta 7 this
week, on Friday. So we may not want to do the db rewrite right away,
unless someone really thinks they can do it in a day, in which case we
might consider it.
Of course, it may turn out that portability or documentation problems
will delay Beta 7 until after I get back from my vacation on August 6th,
in which case we will have plenty of time to convert over to the db
interfaces....
- Ted
