httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rodent of Unusual Size <Ken.C...@Golux.Com>
Subject The way forward: 1.3.7/1.4.0/2.0/2.1/hybrid/mpm/...
Date Thu, 24 Jun 1999 20:47:13 GMT
I've been thinking a bit about the whole release business, and
here's a sort of snapshot of my current opinions.  (Which I
reserve the right to change at any time. :-)

1.3.7 versus 1.4.0:
I don't want to enter into a 1.4.0 development cycle.  I think it
will steal resources from The Next Major Release.  It would require
a beta cycle.  Both of these factors would result in TNMR probably
not even getting to the point of releasing a beta until sometime
next year; would we really want to try to have betas for multiple
versions (1.4 and 2.0) out at once?  I personally don't think so.
I think the only reason we should consider calling the next release
1.4.0 instead of 1.3.7 is if we added something that was a major
functionality enhancement or else made it source-incompatible
with the 1.3 series -- and I personally think that anything that
big ought to go into TNMR.

ISTR the consensus was that the hybrid server would be the basis
of 2.0.  We were going to get 1.3.7 out the door, then synch the
hybrid up with the 1.3.7 tag, and put the result into the apache-2.0
module and start tearing chunks out of it.  I also STR bringing
up my personal belief that we'd be better trying for a single
Major Change per point release, rather than trying to kitchen-sink
all the goodies in the erstwhile 2.0 STATUS wishlist into 2.0.
So 2.0 would be the hybrid (process-/thread-based) server, 2.1
would be I/O layering or some other big thing from the wishlist,
and so on.  I don't recall if anyone agreed with my opinion on
that, though. :-)

Somehow we now seem to have parallel 2.0 efforts going, the hybrid
stuff and the mpm stuff.  The lines are not clear to me.. but I feel
that stuffing the mpm stuff into the 2.0 module when it's so close
to being opened as the main development tree was probably not the
best thing.  I don't really have an opinion on the relative merits
of the various 2.0 efforts, but I *do* wish Dean had held off a
while longer, or put the mpm stuff into the current sandbox (apache-apr)
instead.  Oh, well.

Someone else (Ben H?) has mentioned before that we seem to get really
caught up in the numbers, possibly ascribing too much importance
to them.  I like the idea of not assigning a number until we're
ready to head for the door with something.

(Remember, this is all MHO.)  I'd like to toss out the following
ideas as suggestions toward mapping the way forward from here.
Concerning 1.3.7, though..  As RM, I'm imposing a freeze on feature
addition to 1.3.7-dev; only doc- and bug-fixes, please.  Things in
the STATUS file that have already garnered enough +1 votes get special
treatment: they can go in IFF they won't force mod_perl or mod_php
(or mod_jserv?) to release an update just to get them to build
against us.

Here are the suggestions:

o Close up and release 1.3.7 as quickly as possible.
o Create a couple of sandbox modules with names unrelated to
  any sort of release sequence; maybe a mineral and a vegetable
  (apache-vinyl and apache-oak, possibly?); move the mpm stuff
  into one of these.  Give these *different* CVS mailing lists,
  or none at all.
o Figure out what we want to do re the original plan to make the
  hybrid the basis of 2.0, considering the recent mpm work.  I
  personally would like to stick with the original plan, at least
  for now, because I've got a bunch of stuff that has been waiting
  for the hybrid/1.3.7-merge->2.0 step.  Then Good Stuff from
  the sandbox(es) could be backported into the 2.0 module without
  keeping either from moving forward independently.
o Alexei pointed out that if we don't get a 2.0 out by this December,
  we might as well not bother.  I think that's still reasonable;
  we're not following the 'release early and often' model any more.
  So I'd like to posit a *goal* of a 2.0b1 by the end of August 1999.
  Not a hard dated milestone, just something for which to shoot.

Just a brain-dump.. interrupted by a long telephone call, so
I probably missed something obvious.  Oh, well.
#ken    P-)}

Ken Coar                    <http://Web.Golux.Com/coar/>
Apache Software Foundation  <>
"Apache Server for Dummies" <http://Web.Golux.Com/coar/ASFD/>

View raw message