httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roy T. Fielding" <field...@kiwi.ICS.UCI.EDU>
Subject Re: 1.3.10 release... postponed
Date Tue, 28 Dec 1999 18:32:14 GMT
>THIS TIME IT WAS NOT MY IDEA TO NOW INTEGRATE EAPI INTO 1.3!
>So why to the hell do I find myself here attacked, just
>because I thought consensus was already found by others?

You are not being attacked, Ralf.  It is time for you to stop treating
criticism of your code as if it were a personal attack.  All of us here
are aware of your DSO design criteria for EAPI.  Some people may disagree
with it, but I don't.  I do disagree that the implementation chosen is
the right one given the design criteria, because performance cannot be
ignored.  In all cases, performance is more important than DSO, period.
DSO just isn't that important a criterion that we should make the
primary extensibility framework suck just to support it.  I don't care
about type safeness, but we can't "fix" the performance issue once
the API is set in stone by a release.

I believe that you can obtain all the DSO-safeness you want in an API
by replacing the strings with a config-time loaded string symbol table
and then use the indices within that table as your jump point.  The reason
I haven't implemented such a thing is because I have no immediate need
for EAPI (and far too many needs for other things).  The reason this
isn't in 1.3 is because we don't have enough members working on 1.3
to support a bugfix release, let alone a new feature set.

This is not ping-pong.  Nobody here has changed their minds one iota
since the topic was first discussed.  Just like any new feature, it needs
to earn its way into the base release, and the way it earns it is by
being good enough to be worth maintaining by the group and enough
volunteers signed up to help maintain it.

EAPI is being discussed now because somebody else asked for it to be
in 1.3.  We generally don't ignore people just because the topic has
been discussed before, even though sometimes we should.  The only way
to resolve the topic is to encourage more people to become active in
development so that we can consider working on two branches at the same
time (as opposed to half a branch, which is the current situation).
Actively discouraging people from working on Apache doesn't help with
this situation.

There is absolutely nothing stopping you from working on the 2.0 branch,
including replacing the hook stuff with something that is equally performing
and more functional than what Ben added.  Nothing in 2.0 is sacred.
However, once it is released, it will be sacred (just like any API),
so now is the time to change it.

At the moment I am installing Linux on a laptop so that I can start
developing again, since I'm tired of waiting for my Sun to show up
and there is no way in hell that I'm going to try serious development
under Windows NT.  But even when that happens, I'll be concentrating on
the core server and not the extensibility API, at least until I figure
out how the new API is supposed to work.

....Roy

Mime
View raw message