Re: Getting Struts to work with XYZ

[Timur Mehrvarz <ti...@d2mannesmann.de>]

> Well, one thing you might wish to note is that I'm the primary 
> author of Struts, and my "day job" (what I do to make a living) 
> is developing on Tomcat (in the version 4.x codebase). 
> As you can imagine, Tomcat is my primary test platform
> for debugging Struts, so you can pretty much always count on 
> those two playing nice together.
> Craig McClanahan


Sure, but from your view, how are the rational chances for someone 
to always catch up with the complete functionality (and behaviour) 
of TC x.x (reference implementation) to make something like Struts 
work solid on it?

I get the impression this is a dead race.
Suppose XYZ puts a lot of effort in it and can get their product 
to be 99.9% compatible with TC 3.2. Does that mean they can do it
again later (for 4.x)? I just don't think so.
(sounds as if I'm talking about OS/2)

Would'nt it (for us in need of Struts) be much smarter to avoid  
such problems and setup TC in front of our EJB-container?

By the way, how do you judge the advances of running SE and 
EJB-container in the same VM compared to a) running on seperate 
VM's and b) running on separate machines? 
I can't express how glad I would be to see a discussion of that.

Thanks.
Timur Mervarz

Re: Getting Struts to work with XYZ

["Craig R. McClanahan" <Cr...@eng.sun.com>]

Timur Mehrvarz wrote:

> > Well, one thing you might wish to note is that I'm the primary
> > author of Struts, and my "day job" (what I do to make a living)
> > is developing on Tomcat (in the version 4.x codebase).
> > As you can imagine, Tomcat is my primary test platform
> > for debugging Struts, so you can pretty much always count on
> > those two playing nice together.
> > Craig McClanahan
>
> Sure, but from your view, how are the rational chances for someone
> to always catch up with the complete functionality (and behaviour)
> of TC x.x (reference implementation) to make something like Struts
> work solid on it?
>
> I get the impression this is a dead race.
> Suppose XYZ puts a lot of effort in it and can get their product
> to be 99.9% compatible with TC 3.2. Does that mean they can do it
> again later (for 4.x)? I just don't think so.
> (sounds as if I'm talking about OS/2)
>

Being 99.9% compatible with Tomcat x.x is not the point -- being 100% compatible
with the specs is the point.  Tomcat, like any piece of software, has its own
bugs, but a very large amount of attention is placed on ensuring that Tomcat is
as spec compliant as it can be.

Other platform providers decide for themselves how important that is to them.
Pretty much all the container providers are represented in the expert group that
*defines* those specs, so they don't have a lot of excuse IMHO not implement the
requirements.