On Fri, Jul 16, 2004 at 09:53:11PM +0200, Sander Striker wrote:
> Butting in: it would have been a bit nicer on Dan's toes if a
> simple email was sent: "Remember we are in R-T-C...", instead of
> the instant revert.

Ok, I wanted to write about this since there are some hard feelings out
there right now.  We chatted a little in IRC, but I wanted to put this
up on -dev.

To start, while I absolutely believe the revert was necessary, and needed
to happen before a vote could occur, I do agree I should have spoken
with Dan first about how to proceed.  At the time, I thought we all
had agreed it was a non-trivial change and needed a vote.  As such, the
patch shouldn't have been committed, and as such it should be reverted
(see below for my thoughts about why this is).  Then a quick vote,
reapply the patch, and we'd be on our way again.  Since we all apply
each other's patches from Bugzilla most of the time, I don't think of
the code as something personal.  Therefore, I didn't think the revert
would be considered offensive.  It just didn't occur in my mind.

So Dan, I apologize for not communicating directly with you about this
before touching the code.  I don't know what else I can say regarding
this, so how about the next time we meet I'll buy you a beverage of
choice, and we can put this behind us?




> I'm sure we all agree that people are more important than dry
> procedure ;).

Yes, I agree that people are more important than procedure.  I think we
can all agree, as well, that the people involved in the project are all
trying to "do the right thing" to move everything forward.

However, the fact that there is a procedure that was developed, and
discussed, and voted on, means something.  If there are policies and
procedures they are typically there for a reason and ought to be followed.

I have never worked on an Apache-related project before from a developer
side of things.  So I have only my past experience to go on.  As such,
I have worked in several software and hardware engineering shops
over the years.  At pretty much all of them, the stable "mainline"
(or "HEAD" or "trunk" or whatever the name) is considered, for lack
of a better phrase, "holy ground".  As such, the only code that can
go into the mainline must be "blessed" before it can be committed.

I've seen it be simple (the code compiles), to much more
cumbersome (review by multiple senior engineers, full 25000+
regression test suite run on 12+ platforms, etc.)  For us,
so far, the only real corollary would be our RTC policy set
(http://wiki.apache.org/spamassassin/DevelopmentMode).  The goal of
these policies/procedures is to keep the stable branch stable.

So my reaction, when something happens counter to the policy, is that
we need to revert to a known "stable" point, then continue forward again
as appropriate.

-- 
Randomly Generated Tagline:
"Kluge.net belongs to Theo, my ex-roommate from Worcester, who I can say
 with some measure of admiration, is insane."
                         - Alan Caulkins, http://www.maxint.net/~fatman/

Attachment: pgpHw3fmFPGsJ.pgp
Description: PGP signature

Reply via email to