cocoon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sylvain Wallez <sylv...@apache.org>
Subject Growing, overall consistency and deprecation (was Re: Janino, an embedded java compiler)
Date Fri, 03 Dec 2004 15:22:58 GMT
Niclas Hedhman wrote:

>On Friday 03 December 2004 22:23, Sylvain Wallez wrote:
>  
>
>>Ok, re-read this with a [OT/RT/FS] prefix in the subject. I have not
>>said I *want* this to go into Cocoon.
>>    
>>
>
>Sorry I chose to make the reply on your mail, and not one of the other feature 
>proposals where it perhaps made more sense... 
>
>Just wanted to raise a small flag. The creep is scary because one by one, 
>everything makes sense but looking at the accummulated total, I am seeing a 
>picture that scares people away.
>  
>

I both agree and disagree: Cocoon is a complex beast, we all know that. 
And despite my long term involvement in its development, there are areas 
that are totally obscure to me, and some features I don't like. Even 
some I wrote myself ; we all evolve over time.

However, the current discussion around templates and expressions 
language is not meant to add yet another feature to Cocoon. It is meant 
to provide a single solution for needs that are currently implemented in 
Cocoon, but duplicated in many places with as many variations as there 
are duplications, as each component (or developper) has reimplemented 
the feature without really taking care of an overall consistency of the 
Cocoon system.

Templates and expression languages are system-wide concern, and not only 
concerns of individual high-level components. So we must define some 
system-wide architecture to hold implementations of these concerns. Now 
do we have to sacrifice diversity for this, by imposing e.g. a single 
expression language? We cannot. Cocoon users form a very diverse 
community, using the platform in diverse environments, even for a single 
person. I want a bean-property EL when I use Hibernate, but also a XPath 
EL when I use XML documents. But I don't want JXTG, XSP, CForms and 
others to each either impose a particular EL or each provide a different 
way of choosing the EL. Making a choice should be possible, but in a 
consistent manner.

So yes, new features will be added. But just like CForms (who's still 
using XMLForm?), these new features should become mainstream and 
deprecate their older equivalent.

Now the problem also comes from the combination of a necessary backward 
compatibility that requires us to keep deprecated features and the lack 
of docs that indicate what features are the current mainstream ones and 
those that are only here because of history.

Maybe we should be less shy to deprecate things. But the problem is that 
deprecation means future deletion, which also scares people as it 
doesn't give the image of something stable. Maybe some "mainstream-ness" 
classification would allow people new to Cocoon to more easily find 
their way into the system, while still giving the necessary code and 
documentation to people having "legacy" Cocoon applications.

What do you think?

Sylvain

-- 
Sylvain Wallez                                  Anyware Technologies
http://www.apache.org/~sylvain           http://www.anyware-tech.com
{ XML, Java, Cocoon, OpenSource }*{ Training, Consulting, Projects }


Mime
View raw message