On Thu, 3 May 2001, Andi Gutmans wrote:

>At 12:47 PM 5/3/2001 +0200, Cynic wrote:
>Is it? I'm not that sure. They also release with known bugs although I
>admit to not knowing their exact release cycle so I won't comment more on

Ehem. We release with known bugs too but we don't tell it to anyone.. :)
(of course people can check the bug database..)

>this :) By the way, in Zeev's not very successful way of using that N word
>what he meant was a dictatorship that makes binary decisions about everyone
>else. That's all he meant IMO.

I understood what he meant. Still, I don't think it's a bad idea to have
some kind of committee who decides whether a release is ready to be
released or not. Committee of like.. 10 persons could handle it better
than the current situation where there are 2 persons who decide on
releases.. you and Zeev. Now you guys wait an undefined amount of
time for feedback from QA and if you don't get it you seem to assume
everything is ok. Anyway. I might have misinterpreted this behaviour
of yours but that's how I see it. When we have 10 dedicated individuals
who represent the rest, they discuss whether a release is ok or not.
Vote about it or something. The current situation is horrible as nobody
knows the actual status of testing. My point is that the release should
be decided democraticly but still within limited number of people.

But if having any kind of ORDER here is so hard for some people, we
can forget this and continue with this x.x.x, x.x.xpl1/2/3, x.x.x,
x.x.xpl1/2/3 scheme..

Maybe we should have something like the terrorist cells?
ie. groups of 3 persons who know only about the one above them?

    /    \
   a      b
 / | \   / \
c (d) e e   g

and so on. Where X is the 'release master'.
How this works:

QA people c,d report to a that they have tested RC and it's ok.
A reports to X ok. A only reports IF he get OK from both c and d.
Same goes for B who only reports to X if he/she gets OK from e / f.

Then and only then X decides on release when he get's ok from both
A and B. There could also be more people on each cell. Like 3-5.

This is just a wild idea. Ignore if it's too wild. :)
I might have read too many spy books..but this could be effective..
We could have our 'cells' and every cell 'leader' recruits his/hers
testers.. who could then do the same too..and cell would be active
only if it has enough members..:)

>>4.0.5 took very long to release, and it seems like it could
>>have been released a month ago - pl1 is inevitable anyway.
>I don't see a pl1 happening. We will go right to 4.0.6.

Why would we need a pl? As there really isn't anything that
broken which would need it? (nothing that broken that wasn't
broken before)

>I don't think that the QA team is a joke just because they can't
>necessarily pull the plug on a release. Don't forget that a lot of the QA

Yes it is. As long as there is less than 5 persons who can decide
on this, QA is a joke.

>Anyway, why are we continuing to waste time on this instead of fixing bugs
>and emptying the bugs database.

At last he gets to the point! :)

>4.0.5 was pretty fine and we'll tighten things for 4.0.6. I'm sure that
>after this whole Email exchange more people will be watching the release
>branch and more people will be shouting "STOP". And they will be heard!
>It'll work out OK. Don't worry about it.

PHP 4.0.5-Herzleid, PHP 4.0.6-Herzanfall..


PHP Development Mailing List <http://www.php.net/>
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
To contact the list administrators, e-mail: [EMAIL PROTECTED]

Reply via email to