tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Remy Maucherat <r...@apache.org>
Subject Re: [VOTE] Send trunk to the sandbox
Date Wed, 22 Aug 2007 01:21:36 GMT
William A. Rowe, Jr. wrote:
> Remy Maucherat wrote:
>> Development in "trunk" is not done properly at the moment.
> 
> Back up 1 step; define "proper", with pointers to the
> http://tomcat.apache.org/dev/ documents, if they exist.
> 
> Otherwise you are doing a good job of showing the entire vote is
> really nothing but ad hominem attacks between a few developers,
> and has much less to do with code than personalities.
> 
>> There's a large thread on these issues. At the time, the API design was
>> downright bad. It did improve to some extent (without further debate),
>> but this was all done in trunk without any collaboration.
> 
> You just said you improved it; ergo you collaborated.  You don't need
> dev@ mails to discuss commits until someone disagrees, just as you did.

API changing code was committed without prior discussion, and then 
further modifications were made, all in an official branch (I then 
started working on a sandbox branch myself, which was far more appropriate).

> Your veto in that example said "leave it on trunk".  If you now disagree
> that it can be improved to satisfy your original veto, restate your
> technical veto and just ask for the revisions to be reverted.

Everything was explained in the thread. I found the API to be very 
perfectible and too complex. Iterative simplifications could be made 
without affecting functionality. Some underlying implementation was off 
too, if I remember correctly.

> Of course all code is written without collaboration, unless you are all
> trying to reach a consensus character-by-character on how the source
> code is created.  How it's integrated, or if it is at all, is a matter
> for the project to determine together.

I expect some talking when APIs are concerned, or it should be done in 
the sandbox :| Hence, my requests and the vote to move trunk to the 
sandbox, which I think are legitimate.

> I've actually glanced through the various earlier messages and really
> see only two points of view expressed on the list about any of the actual
> code.  Yours, and Filips.  Which leads me to read the whole debate as
> a turf war over Tomcat.  I can't believe this project would already be
> at the melt-down point again, but here we are.  It's not your personal
> playground, nor is it Filip's personal playground.

(excuse me, but when was the previous meltdown ?)

> Feel free to contradict my opinion with a pointer or two at some technical
> input from other project members, of course.  But I become concerned when
> only two people in a project even grok the technical implications of what
> is in their repository.

Yes, only two persons participate in the debate. I think we fully agree 
on the problem, since Filip proposed to vote on proposals but I believed 
it would be pointless since only two persons seemed to hold informed 
votes (or care, which is a bit the same) about it.

> Some backstory,
> 
> As a frame of reference, the same thing happened at APR/HTTPD over the
> entire concept of buckets and brigades.  Ryan and Greg were at odds over
> the implications.  As svn was properly managed with commits/vetos/reverts,
> the projects were at an impasse with no movement to the next generation
> server.  So, all the committers were invited to a f2f powwow for Greg and
> Ryan to duke it out and thoroughly explain their plan and justification.
> 
> We treated it as a non-vetoable situation.  Neither design was technically
> invalid, it was a preference.  So they had their shootout, and (gasp) even
> came to agreement on the appropriate solution (to the nods of dozens of
> attendees).  More importantly, the other committers who were 'inflicted'
> with the design had a chance to thoroughly understand what it was and why
> it was done that way.
> 
> Back to Tomcat, it sounds like this is an argument of preference over
> technical correctness.  Perhaps November in Atlanta or Hong Kong would
> be a good time for you to sit down, fill in all the other interested
> committers in the exact merits of your preferences, and reach consensus?
> 
> And maybe feathercast the debate highlights and decision :)

Thanks for the info, we're not there yet I think.

Rémy


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org


Mime
View raw message