[Date Index][Thread Index]
[Date Prev][Date Next][Thread Prev][Thread Next]

Re: Discussion about stable/unstable, etc.

On Fri, Sep 15, 2000, Franz Knipp wrote:

> [...]
> I see the following problems in the current WML development:
> [...]
> * the development is done mostly by one person - I'd like to split up the
> work and I'd like also to contribute, but also with the possibility that
> not only one person says, how the syntax will change, which feature will
> be implemented, etc. 
> [...]
> * maybe a developer mailing list would help to do work together - so a
> clear distinction between user and developer would be possible, and also
> more discussion about syntax changes, implementation details, etc. which
> could be boring to the "only users"-group. 
> [...]
> I'd really like to participate in the future development of WML...

I've followed the discussions over the last days and today I feel that
Ishould give my opinion, too. So, for what it's worth:

1. It is reasonable to suggest stable and unstable branches of
   development, of course. How these are implemented in practice is a
   different kind of question.
   The Linux kernel versioning approach ("odd numbers are development
   versions, even numbers are stable release versions") I personally
   dislike for smaller projects like WML. They just confuse people here,
   IMHO. But what would be reasonable is to perform a branch in CVS
   named WML_2_0 and from which the subsequent WML 2.0.x versions are
   released from (starting with 2.0.4). To this branch only bugfixes
   are applied by merging them from the trunk. On the trunk the real
   development takes place. Development versions named WML 2.1aX and
   2.1bX are released from the trunk until WML 2.1.0 occurs (where the
   game begins from the start by branching off WML_2_1). 
   In short, this is the FreeBSD development model where development
   and bugfixing takes place on the trunk only and bugfixes are merged
   onto the stable branch ("MFC" [merge from current] in FreeBSD
   terminology). But I do not want to force Denis or any others into
   this model, although I personally would appreciate it, because I have
   the best experiences with it and one does not loose anything.

2. Creating a dedicated developer mailing for WML I would not
   appreciate, although I never would veto its creation, of course. The
   point is just that experience in other projects showed the following:
   a) mailing lists should be mainly split because of too much traffic,
   not just because another topic sounds nice for a dedicated list; b)
   sw-wml is a lower traffic mailing list and so the traffic is not
   expected to explode in the near future; c) if developer discussions
   are taking place side-by-side on a common mailing list where also
   plain users are subscribed, those users have the chance to better
   monitor the development and recognize heavy decisions. 

   So, IMHO the order should be: 1. start discussions, 2. if really too
   much traffic over a longer time (> 3 months) occurs because of those
   discussions, the list is split in order to reduce the traffic for
   plain users. Or for those who want to draw comparisons: Do you know
   why thousands of USENIX newsgroups exists where only 2 posting per
   month occur? (1x automated posting from the newsgroup creator that
   the group is alive and with the wish that people should post; 1x user
   posting saying "me too").

3. I'm involved in really many Open Source projects, ranging from
   very large ones (FreeBSD), over medium ones (Apache), up to small ones
   (WML, Pth, etc.). But really in all projects, on a regular basis
   people pop up with statements like "I would contribute if..."
   followed by a list of wishes which would help them contributing.
   That's all fine and most wishes are reasonable, except for the fact
   that those people treat their wishes actually as hard requirements
   and at the end do not contribute at all with the final excuse that
   their mentioned "wishes" where not fulfilled in the past.

   So I'm always very sceptical if someone says he really would like to
   contribute if just something in the project is changed. Sure, some
   changes would make his life easier, but they are not a requirement
   for contributing. In the Open Source world, everyone who really wants
   to contribute can contribute - at any time. Sure, without some things
   like Anonymous-CVS access, daily snapshots, etc. it is not such
   convinient to contribute. But their lack is not an excuse for not
   immediately starting contributing.

   So my answer to the above "I'd really like to participate in the future
   development of WML" just can be "Then show us your code and start
   contributing by sending your patches to sw-wml or directly to Denis". If
   you're doing this over a longer time, the existing developer(s) will
   automatically accept you as a strong contributor and start adjusting the
   project environment to also to fulfill your wishes. For instance if
   Denis had to apply two large but great contributed patches per week
   to WML over a time range of two or more months he certainly will
   automatically contact me and say "Ralf, can you please give this
   guy direct access to the WML CVS repository - I trust him, he's a
   great contributor and so he should be able to directly commit his
   changes to WML in the future".

                                       Ralf S. Engelschall
Website META Language (WML)                www.engelschall.com/sw/wml/
Official Support Mailing List                   sw-wml@engelschall.com
Automated List Manager                       majordomo@engelschall.com