httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rodent of Unusual Size <Ken.C...@Golux.Com>
Subject Re: Review then commit.
Date Sun, 28 May 2000 00:36:43 GMT
rbb@covalent.net wrote:
> 
> I would like some feedback here to moving to a review then
> commit mode for 2.0.

My feedback is: -1

> We are getting VERY close to the end of the alpha cycle, and it is
> time to stop treating this like a sandbox finally.

Whoops, hot-button time.

My feedback is: crap.  IMNSHO, commit-then-review should be the
rule until we think we're close to a 2.0.0 release candidate.
Then we can become more cautious and focus on fixing problems and
inconsistencies.  Switching too early will stifle participation.

MSNHO is that the repository should *remain* a sandbox for at
least several betas.

And who says we're getting close to the end of the alpha cycle,
anyway?

> One of the big things that is bothering me, is the sheer amount
> of code that went into the tree last night, with an alpha ready
> to be rolled tomorrow or Monday.

The CTR rule is to pre-post things that might be massive or
controversial.  Last night's commit violated this rule.
However, it was one occurrence, and during an alpha, so I
personally hardly think it constitutes reason to go to RTC.
Anyone could have vetoed it and rolled it back; that's what
CTR means.  If you are concerned about the alpha scheduled,
I don't think anyone would have given you grief if you had
done just that for just that reason.

I'm quite concerned about some of the attitudes I think I've
been seeing about the sanctity of alpha, or even beta, releases.
Get a GRIP, people -- Apache is about folks working on stuff for FUN
and (largely) in their OWN TIME.  If we send out an alpha and
it's broken, it *isn't* a big deal.  If we send out a beta and
it's broken, it STILL isn't a big deal.  If we release an
actual release and it's broken, THAT'S a [relatively] big deal.
We're presumably not working on Apache because we have a deadline
to meet or will get a bonus if 2.0 goes out before 2001.  Are we?
We're not a software house with marketing-imposed deadlines.  You
know what I mean, Ryan -- you worked at IBM.  The development
process perks the software until it's ready to go out.  It's worked
for several years and I don't want to see it change.  If it works,
leave it be.  And it does.

It has been said by one some consider wise: "a foolish consistency
is the hobgoblin of little minds".  Breakage is a fact of betas
and even more of alphas.  Let's not obsess about it.

For those who haven't seen it, I recommend the latest (May 26 2000)
ApacheWeek; the graphs showing implementation of the various
Apache versions is quite edifying.

> but IMO our alpha's need to start becoming VERY stable beasts.

I disagree very, very strongly.

> This means taking some responsability to make sure that we aren't
> breaking things days before we put one out.

Only if you accept your previous premise, which I for one don't.

> For right now, I would like to call off the next alpha, and
> straighten out the code that went in last night (I believe two
> or three showstoppers were introduced, but I am still checking,
> and then I will update STATUS), and after that is done, roll our
> next alpha.

That's cool.
-- 
#ken    P-)}

Ken Coar                    <http://Golux.Com/coar/>
Apache Software Foundation  <http://www.apache.org/>
"Apache Server for Dummies" <http://Apache-Server.Com/>
"Apache Server Unleashed"   <http://ApacheUnleashed.Com/>

Mime
View raw message