Carsten Ziegeler wrote:
Daniel Fagerstrom wrote:
We are chosen as committers as induviduals and not as representants for
our companies. From a community stand point I would say that it is time
to deprecate the SQLTransformer. As a representative for my company I
would rather say: no way, we have tons of code that depend on it. It is
a complicated question, but I don't think that the answer is: I need it
at my work so the rest of you should support it.
It is really hard to tell what is still useful and what not. Now, the
simple example of the SQLTransformer shows this: most of us seem to
agree that it's some legacy component and that flow etc. should be used
instead.
Now, think of a reporting tool done with Cocoon. This fetches some
hundreds of MB out of the database and just displays them. In this case
everything other than the SQLTransformer + Stylesheet is simply overkill
(ok, XSP+ESQL is fine as well) and too memory/time consuming.
Agree. The really ugly part of SQLTransformer is its ability to perform
insert/updates.
I'm using ESQL in a number of places for publication purposes, and many
people agree that ESQL is what keeps XSP alive. We have to build an
equivalent for CTemplates.
With 2.2 we will come to an own release cycle per block, we have defined
apis. So we can simply build the 2.2 version of a block, like the
databases block, and as this block only uses the well defined public
api, it will run for the next years without any changes. And this means
no new releases of this block are requried. So the maintenance for this
block should be near to zero - Ok, I know that this is the theory and
the practice will be slightly different...but anyway, this is one of the
aims we target with 2.2.
And that we more or less currently have. A number of "finished" blocks
(that no more evolve) haven't been touched for ages.
Sylvain
--
Sylvain Wallez Anyware Technologies
http://people.apache.org/~sylvain http://www.anyware-tech.com
Apache Software Foundation Member Research & Technology Director
