httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Justin Erenkrantz <jerenkra...@apache.org>
Subject Re: cvs commit: httpd-2.0 STATUS
Date Fri, 07 Jun 2002 03:37:28 GMT
On Thu, Jun 06, 2002 at 10:29:33PM -0400, Ryan Bloom wrote:
> 
> ++1 to everthing in this e-mail.
> 
> Thank you Greg, you have just stated everything that I have been thinking
> for sometime.  Remember when I tagged 2.0.33?  I did that without any
> notice to prove that it could be done without a four week discussion
> on-list.  Granted, we didn't release that version, but it was better than
> the previous tag, so we could have if we had wanted to.

Oh, both of you can go cry me a river.  If you don't like it, go make
your own release.  And, Greg is right - he could release .37 now.
But, one of the rules that Roy mentioned is that the release must
be made with the idea of public consumption.  What value would a
.37 alpha only released for fixes be over the nightly CVS snapshots?

However, I feel that:

- Poorly tested releases don't fare well at all.  As much as you two
  go screaming that we should blindly take stuff from CVS and call it
  a release and be done with it, I just haven't found that practice
  to work at all.  I'm free to do my own thing as you are.  I just
  don't buy that your strategy leads to high-quality and
  well-coordinated releases.  I have seen this strategy back-fire
  too many times.  But, I obviously can not stop both of you.  Go
  ahead and waste your time - and mine unfortunately since I'll
  have to test your release.

- A showstopper can be removed from STATUS by either majority or by
  fiat of the RM.  Start trying to convince people that it isn't a
  showstopper.  I don't much care what we call it - I'll call it
  broken until we properly fix it.  I'm not the RM and I'm only
  one person.  But, I spent way too much time looking at this code
  and I'm not at all happy with what we're doing right now.

- I personally don't feel the tree is stable enough for a GA release
  right now.  We've been making a lot of changes.  I think these are
  changes that we have to make.  However, have people tested it?
  Have we even tried running the new input filtering code anywhere?
  AFAIK, no.  It passes httpd-test (perhaps), but that's about it.
  Has anyone thought to run it either on icarus or daedalus?  And the
  last time we tried to deploy to icarus, we had to retract it
  because it segfaulted.  I would be much more confident if we were
  to get real-world feedback on this code for a period of several
  days.

- The reason that I would prefer to see module API changes 
  straightened out sooner rather than later is that with each
  subsequent release more and more third-parties are adopting
  httpd-2.0.  And, I firmly believe that we should take the time to
  get right the things we know that are wrong in the module APIs.
  If you don't care about that, fine.  I do.

This is a free world.  We have a set of constraints that we all
abide by here in httpd-land.  We can peacefully co-exist even
though we hold diametric positions on critical matters.  Feel
free to release 2.0.37, but I don't think we're ready for a GA
release.  I'm not interested in releasing alpha releases.  I'm
interested in releasing GA releases.  The only way to do that
is to fix the code and test it extensively.  -- justin

Mime
View raw message