I wonder if we should actually imitate Hibernate in this... Don't they
do runtime compilation of enhanced classes?
--
Ing. Leonardo Quijano Vincenzi
DTQ Software
Mind Bridge wrote:
Hi,
2) Work with mind bridge to see what these patches he had for OGNL
were a
long time ago that theoretically sped up runtime performance of these
operations. I see that these things are already being compiled and
cached
though, so I'm wondering if this effort is even needed.
I am afraid standard compilation is not sufficient. Even compiled,
OGNL uses reflection to access the data. In the optimization I used
Javassist to access the data directly with compiled code to JVM, which
resulted in the significant speedup against compiled OGNL code (9
times, as far as I remember). This did not cover the whole OGNL, but
only property access, however.
Drew started working on OGNL 3.0 which I believe was designed to
handle such "compilation" (instead of reflection). If anyone desires
to do so, they could look at the code there and determine its
potential...
-mb
Jesse Kuhnert wrote:
If there was a theoretical person who wanted to cut their teeth on
tapestry
a little more closely than just using it, what would everyone like
them to
work on ?
I know we'd all like documentation / various infrastructure stuff
done but
this is more an exercise in learning tapestry internals as well as
proving
coding abilities and such...So, the value has to be on both sides.
I don't think there's anything we could throw at this theoretical person
that they couldn't handle, so feel free to think of whatever you
like. Keep
in mind that I think we should probably just pick one for now, so think
about it before voting :)
My two ideas so far are:
1) Find a way to do what I described earlier with elimintating the
need to
specify a translator binding...If this means working with Patrick
Lightbody
over at opensymphony (where the ognl code now lies dormant) , then so
be it.
2) Work with mind bridge to see what these patches he had for OGNL
were a
long time ago that theoretically sped up runtime performance of these
operations. I see that these things are already being compiled and
cached
though, so I'm wondering if this effort is even needed.
3) Work with whoever exists in the community doing hibernate things with
tapestry and make Howard's EditObject blog entry a reality. This would
hopefully be done in some project either hosted as a sub project on
http://howardlewisship.com or as part of an existing project, like
honeycomb
/ James Carmen's work with trails / hiveutils / etc...
I'm personally hoping for #3 because I think it will provide the most
value
for the community. Everyone's input is welcome.
--
Jesse Kuhnert
Tacos/Tapestry, team member/developer
Open source based consulting work centered around
dojo/tapestry/tacos/hivemind. http://opennotion.com
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
