From: "Randy Terbush" <randy@covalent.net>
To: <new-httpd@apache.org>
Subject: RE: am I dreaming?
Date: Thu, 25 Mar 1999 13:38:13 -0600
Message-ID: <007301be76f7$057b02e0$0502a8c0@covalent.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="Windows-1252"
Content-Transfer-Encoding: 7bit
In-Reply-To: <199903241251.HAA17215@smtp3.mindspring.com>
Importance: Normal
Sender: new-httpd-owner@apache.org
Precedence: bulk
Reply-To: new-httpd@apache.org

I suggested a few weeks back (or was it months) that we should not
consider a mismatch in MODULE_MAGIC_NUMBER to be fatal. This decision
really should be left to the module vendor. The module should know
what range of MAGIC_NUMBER it can accept and decide to participate as
a registered handler or not. That simple change would solve a lot of
these issues.

I still think that little changes like the one desribed should be
MINOR number changes. That was my intent when I added
MODULE_MAGIC_NUMBER_MINOR.

If we can agree on a way to deal with this, I would be happy to work
on it.

-Randy

> -----Original Message-----
> From: new-httpd-owner@apache.org
> [mailto:new-httpd-owner@apache.org]On
> Behalf Of Vince Bonfanti
> Sent: Wednesday, March 24, 1999 6:54 AM
> To: new-httpd@apache.org
> Subject: Re: am I dreaming?
>
>
> On 3/23/99 11:21 PM, Thomas Reilly wrote:
> >
> >After a more diligent scanning I noticed that the
> >MODULE_MAGIC_NUMBER_MAJOR was changed because METHODS and M_INVALID
> >were redefined.  Maybe these changes were important bug
> fixes but they
> >seem to just be cosmetic to me, in which case I feel like
> they could
> >have been deferred to a later Apache version.
> >
> >Is backwards compatibility of precompiled DSO modules not something
> >that Apache aims to maintain?  It seems like a very
> desirable feature
> >to me but the fact there have been 20 changes to
> >MODULE_MAGIC_NUMBER_MAJOR since 1.3 left beta leads me to
> believe that
> >I'm the only one that feels that way.
> >
> >--
> >Tom Reilly
> >Live Software, Inc
> >http://www.livesoftware.com
> >
>
> We also "feel the pain" of this lack of compatibility,
> though perhaps not
> as strongly. We distribute source for our product on UNIX platforms
> because recompiling is a relatively painless process for
> the end user.
>
> On Window NT it's a different story: most end users don't
> have the tools
> installed to compile anything, and if they did, the
> complexity of the
> build process is anything but trivial. Therefore, we have to ship
> binaries for Windows NT, and have had to to build a new
> version for each
> Apache dot release.
>
> We would also like to see backwards compatibility of
> precompiled DSO
> modules a higher priority.
>
> Vince Bonfanti
> New Atlanta Communications, LLC
> http://www.newatlanta.com/
>