On Fri, 14 Sep 2001, Pier Fumagalli wrote:
> Anyway, switching to APR is not that easy (IMO), if you want to take
> advantage of ALL which is provided by APR (I'm deprecating ANSI-C here).
This is an incremental process, and can only increase the stability of jk.
> > Refactoring/cleaning of jk - one part will be done by the move to APR
> > ( of course ), there are some optimizations and improvements in the jni
> > connector I am planning ( also after APR is in ), and some of the webapp
> > API ( and docs :-) could also help a lot.
>
> DOH! I'm saying, let's refactor WebApp how you guys want it... In the past 6
> months I got to know APR quite well, on the other hand, you guys have more
> experience on the different web-servers... We have a working APR-based
> implementation, let's just put it together in a nice way so that it fits
> BOTH needs...
It doesn't make too much sense. You're saying we should use WebApp because
it already uses APR, and port back all the protocols and modules to it.
I don't know - if using APR is so difficult that it's easier to port all
the modules to webapp rather than replace the C functions with the APR
equivalent in jk - then maybe we should thing again about using APR.
But my impression so far is that APR is quite easy to use and very close
to the current abstractions used in jk.
Costin
