On 4/24/2010 3:06 PM, Jeff Mitchell wrote:
On Sat, Apr 24, 2010 at 12:20 AM, Victor Duchovni
<victor.ducho...@morganstanley.com> wrote:
On Sat, Apr 24, 2010 at 12:01:05AM -0400, Jeff Mitchell wrote:
On Thu, Apr 22, 2010 at 6:44 PM, The Doctor<doc...@doctor.nl2k.ab.ca> wrote:
??Out: 220 doctor.nl2k.ab.ca ESMTP Postfix (2.8-20100323)
I know this (probably) has little bearing on the problem at hand, but
if I used experimental snapshot releases in a customer-facing setting
I'd be booted out the door.
Postfix 2.8 snapshots are reliable enough for production use. Their
feature-set is not future-proof, but if one needs (say postscreen)
a snapshot feature, it is not unreasonable to use a tested snapshot.
The terms "stable" and "experimental" (and "alpha" and, once upon a
time, "beta") have fairly widely understood meanings conveyed to users
by software developers. Experimental software may be reliable, but
there's a reason that the developers have not yet marked it as
"stable" -- when they feel the time is right, they will do so. There
are of course always reasons why you may want to disregard this
labeling (such as needing new in-testing features this moment).
This isn't a dig at Postfix -- I'm sure the snapshots are quite
reliable, and I've never had anything but rock-solid, reliable
behavior from Postfix. But I have certainly seen regressions in
testing/experimental releases with other programs with rock-solid
stable releases. As a policy it generally makes sense to listen to the
developers' labeling unless there are specific reasons otherwise...
--Jeff
Then read the developer's labeling.
"Stable releases do not change except for bugfixes and for
portability fixes. New features are tested out in experimental
releases"
"New features are tested in experimental releases. They become
part of the next official release once the code has not
changed for a significant amount of time. Although this code
is still subject to change, it runs on all of Wietse's systems
so it is production quality."
