httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jim Jagielski <...@jaguNET.com>
Subject SGI's Accelerating Apache Project and contributions
Date Thu, 22 Feb 2001 03:46:29 GMT
First of all, this is coming from someone who is personally upset
and saddened by the "demise" of the project.  I considered it a worth-
while and useful project, and would frequently mention it (in fact,
I did so in my Jan 2001 Web Techniques article).  But I am also
a member of the ASF, so this is from someone who is wearing both
hats at the same time.

There has been some "discussion" and assumptions being bandied about
since the news that the SGI's Accelerating Apache Project was being
dropped.  It's been implied (and stated at times) that it's due to the
ASF either ignoring the patches or else the ASF is a "good old boy's
club" and was preventing the patches from being included.  The truth is
much more simple.

When the patches were originally sent, they were sent in a combination
of sequential patches.  The second depended on the first, the third on
the second, and so on.  So that the tenth patch relied on the first nine
being applied or it wouldn't apply.

The first patch had a bunch of casts and other constructs that we have
tried very hard to remove from, and keep out of, the Apache code because
they are indications of somewhat sloppy programming and can lead to very
subtle bugs.  When the patch was first sent, we replied basically saying
as much in an attempt to help get the patch into the code.  The patch
required a lot of small modifications to get it into our coding style
and standards (sometimes called "the apache style"), so that it could be
committed.  Those modifications were never done, by anybody.  Because
every other patch relied on that first patch, none of them ever got
applied. It's the work of the patch submitter to "adjust" the patch to
the accepted standards of the official code base.  Yes, this means that
there are rules and standards that the contributor needs to abide by,
ignoring them isn't an option.

The 2.0 patches are a different matter.  The Apache Web Server Group
believes very strongly in the BSD license.  The State Threads MPM that
Mike has created uses NSPR (under the MPL 1.1) for some of its
functions. We just won't accept any patches to our code that are
contributed as MPL 1.1. So the State Threads MPM must be left out of the
tree.

There is also the very real fact that just because patches are
submitted, it does not mean that they *must* be included in Apache. Code
is committed when it meets the known requirements of what it takes. It's
really bad form for someone to say "The ASF didn't accept my patches, so
they should be forced to."  This is not how open source works, and it's
certainly not a recipe on how to produce good code.

We did try to work with Mike to get those patches in to the code, and we
honestly believe it was a failure of communication more than anything
else that kept the code out of the tree through this present day.  And
this will continue to be a roadblock, until the AAP coders actively
participate in discussion of their patches and work *with* the ASF to
everyone's (including SGI's) satisfaction.
-- 
===========================================================================
   Jim Jagielski   [|]   jim@jaguNET.com   [|]   http://www.jaguNET.com/
          "Casanova will have many weapons; To beat him you will
              have to have more than forks and flatulence."

Mime
View raw message