Hi Jacek,
it sounds like what you are doing is a branch
of the XSLTC codebase.
If the changes you are making are that radical,
they should really, *really* be done as a CVS
branch in the xml.apache.org CVS server. This
has no destabilizing effects on the Xalan
baseline. Only when the code on the branch is
stable and shows marked improvements will it
be merged onto the trunk. Meanwhile, changes
occurring on the trunk that are appropriate
to the branched code as well can be merged as
you go along, to ease the final merge onto
the trunk. This is easy, does not compromize
stability, allows others to easily work with
you on the experimental XSLTC changes, and
promotes easier future use of those changes
in the baseline XSLTC product (pr
The alternative, which is effectively what
you are doing now, is to spawn a new
(presumably open-source) project out of
the XSLTC codebase. Obvious flaws in that
approach:
- it fragments the development effort
- it doubles the open-source infrastructure
(assuming you will allow others to make
changes in your forked project) which is
unneeded effort -- there is an established,
secure CVS server already running, a bug
database and all that -- why not use it? :)
- the longer this goes on, the harder it is
to eventually merge the two versions ...
if the two branches aren't in the same
CVS, they will grow apart.
So for the benefit of XSLTC, I strongly
urge you to reconsider! :)
Cheers,
- Gulli
> -----Original Message-----
> From: Jacek R. Ambroziak [mailto:[EMAIL PROTECTED]]
> Sent: 6. mars 2002 18:17
> To: [EMAIL PROTECTED]; [EMAIL PROTECTED]
> Cc: [EMAIL PROTECTED]
> Subject: experimental XSLTC-2, a clarification
>
>
> Dear XSLTC users,
>
> I owe you clarification about the status of
> my XSLTC version, binaries of which are served
> from my website www.ambrosoft.com.
> The version is derived from Apache/Xalan/XSLTC 2.3.1
> which in turn is derived from my final project
> at Sun Microsystems.
>
> I have returned to this topic because I myself need
> XSLTC for my commercial work; the direct trigger was
> an email from Stefano Mazzocchi on lack of benchmarking
> data and bugs prohibiting him from the unrestricted use
> of this promising technology in Cocoon.
>
> I started to work on XSLTC again after a long break
> thinking that perhaps it was a matter of fixing a few
> outstanding bugs and everything will be fine.
> After many long days of intensive effort I realized
> that the code is in fact quite far from fully
> realizing the potential of the original vision.
> This is my personal opinion. I am not going to discuss
> details here, but improvement opportunities are numerous
> and I am implementing them.
>
> My current version represents an uncompromising overhaul
> of the code base and is no longer implementation compatible
> with Xalan/XSLTC (although it is API compatible).
> Therefore at this point I cannot CVS-commit nor -update
> as doing so would mess up the Apache and/or my versions.
>
> I am calling my version "experimental" because I indeed
> experiment freely with any, however far-reaching, changes
> I feel are beneficial to the technology. I am addressing
> performance and architectural issues with a fresh perspective.
> Maybe my long break was not such a bad thing after all :-)
>
> This level of free experimenting would not be possible
> in the context of the Xalan community effort,
> neither would it be a good thing: it would be destabilizing.
>
> Naturally, I wouldn't be investing energy into
> reenginering XSLTC if I didn't believe a worthwhile outcome
> would result.
>
> I will continue to post information about XSLTC-2
> at my website, but I strongly feel it inaproppriate to
> continue discussing it at this forum. That's why I had asked
> people interested in progress/use of XSLTC-2
> to contact me directly.
>
> I am sorry if this explanation disappoints some of you
> but I hope my results will end up being useful
> to XML/Java users.
>
> --Jacek
>
>
>
> _________________________________________________________
> Do You Yahoo!?
> Get your free @yahoo.com address at http://mail.yahoo.com
>
