httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jim Jagielski <>
Subject Re: Starting over..
Date Fri, 12 Sep 1997 16:28:11 GMT
Roy T. Fielding wrote:
> Please folks, don't make the same mistake twice.

Don't worry... I won't. I had almost stepped up again
to be release manager for 1.3b1. Thanks for knocking some
sense into me.

> There is absolutely
> no point in setting a date as a deadline for getting something done.
> We have no boss, no marketing/sales pressure, no external calendars
> that need to be synchronized with suppliers, and no quantifiable
> estimate of the rate at which we as volunteers will get things done.
> It is therefore wrong to attempt to "manage" this group as if it were
> a normal software development effort.

In others words, since we are not a "normal software development
effort" it is wrong to try to avoid the problems that were
clear in the delay in getting 1.2 out. It's "wrong" for someone
to say, "hey, now's a good time to maybe put out the beta for
1.3 so we can get back on track, get the NT port out there and then
start working on 2.0". Instead, we'll all wander around the field
like cows until, by random chance, we all end up in the barn.
Moo :)

Nonetheless, if this is true, then someone should delete the
Apache Server Project Plan, since this sets dates as well (dates
which are woefully wrong).

> Consider how I was running the Agenda.  We had a list of showstopper
> bugs (things that had to be fixed before the next release), a list of
> current patches (things that people had completed work on), and a
> list of other known bugs that were within the realm of being fixable
> before the final release.  I wouldn't start a "countdown" toward a
> release until all of the showstopper bugs were fixed and all of the
> current patches either applied or rejected.  In other words, the
> "deadline" was an identifiable goal that everyone agreed upon --- removing
> the important items from the agenda --- and didn't have to be renegotiated
> every time a new problem was found or a useful patch was proposed.

AFAIK the status reports did report these things. Everyone got those
reports so, if there were ommisions or mistakes, they could be
corrected. If there were show-stoppers then people would hopefully
let the status-guy know. The schedule slipped a few times because
we weren't ready and that's cool and we can do that because
we aren't normal. But having a goal to shoot for is good too. I
don't think saying "let's try for a release on such-and-such a date"
is necesarily a bad thing.
> That was more than sufficient to motivate a release.  Quite frankly,
> I am astonished that people want to push for a release of 1.3b1 when
> there has been scant evidence that the current cvs snapshot even
> works on NT.  I suggest that a list of goals be written into the
> agenda, and that we tarball 1.3b1 as soon as those goals are achieved.

One condition has always been that we don't release 1.3b1 until
the NT people say it's ready. Since, in my mind, the whole
focus of 1.3 has been the NT stuff, that should be the priority.
We were all excited about having a port of Apache for NT, so
much so that we put out an ALPHA release. We did this because of
a demand and desire for Apache on NT. If the NT people think that
the code is ready for a beta release, then I think we should
release it.

In any case, I'm not falling into this trap again... It seems there
is both great desire for a beta, great resistance, and great "I
don't care". Damned if you do, damned if you don't.

I'll pass thank you. :)
      Jim Jagielski            |       jaguNET Access Services           |
            "Look at me! I'm wearing a cardboard belt!"

View raw message