You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Ian Holsman <li...@holsman.net> on 2002/09/15 05:53:36 UTC

Magic Number bump (was Re: mod_custom_log exits too late?)

On Sat, 14 Sep 2002 13:00:55 -0700, Aaron Bannert wrote:

> On Sat, Sep 14, 2002 at 07:16:36PM -0700, Ian Holsman wrote:
>> > Yes, Huge +1!!
>> > 
>> I disagree on this.
>> if we know we are about to break something then we should do it >before<
>> we release.
> 
> Although this sounds tempting, it it just another "me too" that would
> better be ignored by the RM*, **.
> 
> *It's always up to the RM.
> 
> **I'd like to volunteer for to be the RM for 2.0.42.
> 
>> do you realize how many people HATE having to recompile their modules,
>> and what a pain it is for commerical vendors to support?
> 
> I can only think of two times where an admin _must_ recompile their
> modules:
> 
> 1) We made a security-fix release, and they have to be up to date or
>    risk exploit exposure.
> 
> 2) They want to use a new feature that is only in the new version but
>    in the mean time some MMN was bumped.

or 3) the 3rd party module vendors.
&
   4) the win32 / binary downloaders

now the 3rd party module vendors have to provide >3< versions of their
modules (40, 41, and for 42 when it is released)

and all the support emails to their newsgroups/mailing lists saying 
"I just upgraded and your module doesn't work" which I have seen too many
times.

how patient do you thing the 3rd party developers are? do you think they
are going to continue to rebuild/fix their stuff just because we have a
release cycle which seems to be a release every month? or do you think
they will just not support .41 and recommend users to stay with an older
version

--Ian

> 
> I have sympathy for the first group. The second group is going to be
> happier that our stuff gets released sooner rather than later, and they'll
> also be happier if the end result is stability. (And to show their
> happiness they'll even volunteer to test our stuff and give us valuable
> feedback). Whenever we hold up a release for a last-minute fix we are not
> only delaying a release (that people are waiting for) but we are also
> increasing the chance that there will be a bug in the release.
> 
> Anyway, this is all IMHO, :)
> -aaron