httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William A. Rowe, Jr." <wr...@apache.org>
Subject RE: Auth: Start the httpd-2.1 branch finally?
Date Sun, 13 Oct 2002 00:17:14 GMT
At 04:05 PM 10/12/2002, Sander Striker wrote:
>> From: Aaron Bannert [mailto:aaron@clove.org]
>> Sent: 12 October 2002 22:18
>
>> On Sat, Oct 12, 2002 at 10:37:07AM -0700, Justin Erenkrantz wrote:
>> > >I don't have a strong opinion about the authn redesign,
>> > >but I do have one change in mind that would fit well in
>> > >2.1: async write support.  And async read support, but
>> > >that may take a lot longer.
>> > 
>> > My belief is that you should design and code up the async support and 
>> > then we can deliberate about where it should go.
>> 
>> That seems like a one-way street to me. How come it's ok to work on the
>> auth changes in 2.0 but it's not ok for others?
>
>The auth changes were complete before they were applied in the sense that
>they didn't leave the tree in a broken state.

But this is exactly my complaint with auth.  It has STILL left the docs, 
and therefore our users in a broken state.  We've rolled two releases 
without what is otherwise good code, because of the impact on users.
In fact, we agreed that the project (we were talking about code) could
and should be broken a few weeks, maybe a month, for major changes.
It's getting on two months and still there are next to no docs for this
change.  So it's totally unreleaseable.

EVEN if the docs are done, how can they help both our 2.0-pre-auth 
and 2.0-post-auth users at the same time???  I'm objecting from a 
DOCS point of view, folks still lookup 1.3 docs on our website!
And from the CONFIG point of view ... folks need their hands held
through this upgrade.  Who has it in their plans to answer all the
bugzilla reports and redundant questions on the users@httpd lists?

It's ok to work on ANYTHING.  We are all agreeing to this.  The question
Ryan raised is, does it belong in 2.0 or 2.1.  As Jim asked, are you looking
for a playground for good ideas or do you have solid problems to solve?
I'm suggesting that 2.1 should exist today.  It took Ryan (and others) over
two years to create a GA tree.  If we continue with design-by-committee,
it's time to begin development.

>> > These changes shouldn't be held up by the fact that 
>> > we don't have a 2.1 yet.
>> 
>> I agree, as a matter of fact, I don't think any changes should be held
>> up for any reason whatsoever.

We all agree.  If we need 2.1 let's create it already.  Nobody ever said that
1.1 was 'complete' before 1.2 development began.  Nobody has ever implied
that the final subversion of any revision is "IT".  It's never "DONE".  Let's
take big changes and call them version bumps when we should.  And get
that to release state and out the door as quickly as possible.

How's this for simple?  Create httpd-2.1 and back out Justin's changes
from httpd-2.0 so we don't break our users.  If someone wants to change
more APIs in httpd-2.1, let them do so.  When it's ready, we release it and
start supporting it just as httpd-2.0.  We effectively drop httpd-2.0 at the
release of httpd-2.1 except for security patches, and anything anybody
really wants to commit.  But the focus is on the last GA code, just as
today.  {Sure little bugs get fixed occasionally in apache-1.3.  Nothing's
wrong with that, or with continuing that tradition in httpd-2.0.}

What's defined as 'ready' for httpd-2.1?  That it works, that it is a GA
quality release.  If we can fix other foobars we made while designing 2.0,
that would be terrific.  Bugs fixed in the httpd-2.0 tree can be committed
to the httpd-2.1 tree.  But let's set our sights on an early release, some
time this winter if not by year end.

Perhaps what scares some developers is the HUGE time between the
idea of 2.0 and it's eventual GA release.  There is no reason we should
get bogged down agian.  Sure, there is a quick alpha-beta-GA cycle, 
but we have that down to a science.

Bill


Mime
View raw message