Hello everyone
as some know, I have been working semi silently the last months in my
opensource time on a jsf dojo layer which is rather extensive, it is a
thin layer on top of dojo currently encapsulating around 23-25 of the
existing dijit components
(around 98% of the dijit components)
I am reaching the state of being able to opensource the codebase, it
still is not done yet, but the core is rather stable and a lot of
components work well enough already so that I am not ashamed of it
anymore showing it publicly although it is still major road ahead till beta.
One thing, it was planned as tomahawk extension to get dojo out of
tomahawk, but it is very extensive and the tomahawk dependencies are not
that much so getting tomahawk out entirely would be possible.
Following:
a) It uses weblets and tomahawk for resource loading and relocation
(weblets for resource loading, tomahawk for resource relocation)
b) It has its own codegen, which was derived out of the need of having
to have something fast, and the old ones were very undocumented and the
new ones not ready. The codegens basically build up a data structure in
groovy with the component descriptors and then send the data structures
down into specialized velocity templates. I know we now have a working
codegen infrastructure, but moving all this over is not my biggest
priority since my own self rolled parserless solution works very well
for the scope of the complib and the component descriptors are very tight.
As for jsf currently JSF 1.1 is still the sourcebase, although I test it
with JSF 1.2, a direct adjustment towards JSF 1.2 can be done thanks to
the codegens.
Have in mind the codebase I am dropping down is rather huge, with around
30.000 - 40.000 locs (Excluding the table) of the components alone most
of the code being generated, and for now also 1-2 dojo versions in the
sourcebase until we can move the dojo builds into a separate maven repo
and load it during the build process.
Also have in mind that currently I drop all generated code into the
java/src hierarchy in its own dirs with clear comments in the files
that they should not be added! An approach which would make sense in my
opinion but others have opposed in the past (I simply love to have the
sources in one place as much as possible)
Also the table which will be added also will have a large codebase, the
customer we have developed the table for, has cleared all the apache
legal stuff, and it is just a matter of getting clearance of the code
from the legal departement to be able to add it.
A demo of the components will be shown in 1-2 weeks, some people in the
list have seen some of the components in action already last week.
So my question is, are we going to host it inside of myfaces as its own
subproject or as part of the sandbox or maybe I can move the codebase
over to its own project outside of apache (jsfcomp for instance might be
a perfect place until the entire complib is matured enough)
