httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William A. Rowe, Jr." <>
Subject Re: Generic cache architecture
Date Wed, 03 May 2006 23:08:27 GMT
Ruediger Pluem wrote:
> On 05/03/2006 11:27 PM, William A. Rowe, Jr. wrote:
>>Moreso, we need more third party authors to -participate- in telling us what
>>in HTTPD-2.4 will make their module better.  And a faster cycle of 6mos-1yr
>>gives them a chance to do this and realize the benefits in the official
>>release more quickly.
> But some of them will fall off the shelf (I guess especially some commercial
> ones), because they do not want to invest time again into an changing API.

Of course.  Why stall all efforts because some don't move with new technologies?

> This means that they stick with an older version and do not do releases for
> each major version, but lets say for every second.

Ok, their user's loss, not an httpd development issue.

Let's be clear here, this isn't a statement against making their lives easier
to actually do the ports - we could be a heck of a lot more helpful in module
authoring and especially in porting documetation.

> Furthermore having frequent major releases increases the backport requests from
> user side and thus likely the backport efforts, as some users stick with older
> versions for whatever reasons (e.g. third party modules).

Wrong.  We honor fewer backport requests for features.  We might entertain some
for bug fixes certainly.  But the answer's always upgrade asap for fixes, and
an upgrade's manditory for new features.

Backporting features is a vicious cycle.  No problem here with folks using
Apache 1.3 when it solves their pain.  But if they want cool feature X and
we give it to them, we support 1.3 for that much longer because this-bug
and that-bug aught to get fixed, and new feature X has a bug so we have
a subsequent release, followed by 12 more pleas for other features to be

Nip it in the bud, freeze the features in old version, and poof, people move
because they *want* the new features, it solves more of their pain, and so they
have an incentive to make *their* investment of time in migrating.  Take away
the incentive and they will not (heck, should not) migrate up to our supported

> Maybe we should have a FDT (Frequently Discussed Topics) to collect the arguments ;-).


View raw message