[ On Thursday, November 16, 2000 at 12:01:21 (-0500), Andrew Brown wrote: ]
> Subject: Re: change in bind?
>
> > What about bind 9.0.0.1?  Is there any reason not to use it?
> 
> * it's still not as featureful as v8

Well, that's a matter of perspective I think.

The "named" in v9 is *far* more featureful than the one in v8.  It's
mostly IPv6 compatible for one, and my favourite new feature is "Views".

The resolver library in v9 is still very lacking in features though and
introduces a whole new set of concepts and APIs.

> * it's not as mature as v8

Given the nature of the code in each I'd say it's more "mature", but not
more "senior"!  :-)

The v8 code is (though greatly cleaned up from 4.9.x) still a mess of
overly maintained spaghetti.  The difficulty with outsiders and
part-time developers such as myself finding the correct fix to the
recent ZXFR DoS bug is an excellent example of just what kind of state
it's in.

The v9 code is a fresh and new and based on a far more "mature" design.
The code might be considered as just barely past the "virgin" state, but
it's already proven to be stable and reliable in several punishing
environments.

> * it currently requires threads

The ISC developers build it with unproven-pthreads-0.17 on NetBSD-current.

So, yes it is a threaded application, but it works just fine with a
pthreads compatible user-level threading library on NetBSD.

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>