OpenACS Road Map
talli says:
hey
talli says:
i was wondering if you would like to write an article with me about oacs
talli says:
... called ...
talli says:
OpenACS and Relational web programming or How I learned to stop worrying and love AOLserver
John says:
d'oh!
talli says:
It will be about why the openacs is not written in an oop framework
and how aolserver provides code reuse
John says:
I was just sitting down to pen a bboard article about why we should
ditch AOLServer
talli says:
holy
talli says:
how come?
John says:
OTOH, writing an article on why it's better than oop would be child's play
John says:
AOLServer is why the framework is not portable
John says:
If it was just TCL/FastCGI it could run on IIS/Apache/freaking Zeus, even behind Zope
talli says:
Actually, don's post suggest that it may be
John says:
and it would be no big deal to integrate things like Java/Jacl, IncrTCL, XOTcl, etc
talli says:
he mentioned that Jamie is able to get it to run on MS because it
was originally written for windows
John says:
But that's such a qualified definition of portable (!)
John says:
*If* Jamie has time
John says:
then maybe it'll run on Windows and be stable
John says:
But it's unlikely he'll be able to keep up with the AOLServer team
John says:
who'll just keep adding features and moving ahead
John says:
If the core team isn't interested in keeping Windows code around,
it's time to think about alternatives
talli says:
true
John says:
for those of us whose client's want to run it somewhere besides *nix
talli says:
that might be true
talli says:
but the reason that the aolserver team is ditching windows support
holds for the oacs community too
talli says:
the vast majority don't want to run on windows, so why support it?
John says:
It's a somewhat self-fulfilling prophecy
John says:
My argument is if you want to move (even slightly) toward critical mass,
you have to bring the non-believers into your tent
John says:
This means the vast majority of corporate Windows shops in the world
John says:
The payback is having a big huge money'd audience
John says:
mo-$$$ for service firms like Musea, for example
John says:
TCL's already a hard sell, but running just the TCL code
w/FastCGI, behind *any* web server, on any OS, with Pg, Oracle, or
MSSQL Server... it starts to become pretty compelling, IMHO
John says:
It's already compelling, but not at all for MS shops
talli says:
so, fastcgi *is* compelling or is it a waste of time?
John says:
I think it wouldn't be too much work. Less than the work I've
already done on the kernel MSSQL port.
John says:
WebWare (a python framework) runs as cgi, fastcgi, or as an apache
module mod_webware
John says:
It's the same code base, you can just swap the listeners in and out
John says:
I think that's the way to go.
John says:
The core code doesn't get touched
John says:
The add-ons (OKI, Java, perl, LDAP, ASP.NET etc) become way
easier for those people with fringe integration scenarios... no more
waiting for one of the 15 AOLServer C hackers to whip up the ns_
module. You start to leverage all the people putting work into the
TCL scripting framework
talli says:
but even with fastcgi, you still need to run aolserver
John says:
No
John says:
Why?
talli says:
because aolserver is the app server more than weblistener
John says:
Meaning it has db pooling, and an API for filtering, and a way to
schedule stuff, right?
talli says:
and what abotu all the tcl apis it requires?
John says:
see nstcl
John says:
http://michael.cleverly.com/
John says:
HAH!
talli says:
ok, well you should post that stuff to the boards
talli says:
see what people say...
John says:
k
8:34:30 PM 
|
|
