David Graham et al. 

I can feel the train derailing off the tracks. Screeeeeeeeeeeeeeeeech!
For a moment let's pretend that I never mentioned BSF or Rhino. I want to
get your collective thoughts on JSTL EL Validator Rule.

JSTL EL Validator Rule (validatejstl) ------------------

What feedback do you have for the JSTL EL validator rule?

David mentioned some disdain for Rhino/JavaScript. What about using JSTL EL
to compare fields in a form for a rule?

How do you think this approach compares to requiredif and validatewhen?

OPTION REQUIREDIF
Is it better to create a complex and verbose set of options like requiredif
to get rid of reliance on an expression language? (IMO very few folks use
this because it is too complex)

OPTION VALIDATEWHEN
Is it better to create a custom expression language just for validation like
validatewhen? 

Would you prefer requiredif, validatewhen, or validatejstl to be part of the
next version of Struts? Or none of them?

Last I heard validatewhen will be part of the next release of Struts. I
think this JSTL EL Validator rule should be included instead. Creating
another expression language for this problem space does not make sense when
a perfectly good expression language like JSTL EL is available. It does not
make sense to maintain a second expression language when Jakarta already has
JSTL, which is better and more mature. JSTL EL is pervasive, e.g., JSTL EL
tags are recommended over legacy struts tags.

I've received some feedback from a few struts users (clients and peers).
They love the idea of JSTL EL validator rule. It just makes so much sense
and solves the problem of doing validation that involves relationships
between form fields. This is a hole in the validator framework that gets
filled quite nicely by validatejstl. The beauty of this is it is so small.

Okay... I am changing tracks to a different conversation.




BSF Validator Rule --------------------

This is a separate issue from the JSTL EL Validator Rule. Perhaps this could
be a separate conversation. 

Having a validate-bsf rule would not necessitate running Rhino... it would
merely be an option.  The validator framework would never have to start
Rhino (JavaScript), Jython (Python), JRuby (Ruby), or any scripting language
unless the web application developer wanted to use this feature.

I like the thought of Rhino because I could create scripts on the client and
the server, and only have one set of validation scripts. I think this is a
better alternative than having to maintain two sets of code that does the
same thing in two languages. Again, this would be an option, not required.

** More Comments below...

Rick Hightower
Chief Technology Officer
Trivera Technologies
http://www.triveratech.com
520 290 6855 (Phone)
520 977 8605 (Mobile)
 

-----Original Message-----
From: David Graham [mailto:[EMAIL PROTECTED] 
Sent: Tuesday, August 26, 2003 6:51 AM
To: Struts Developers List
Subject: RE: JSTL EL Validator rule: A better requiredif and validatewhen
using JSTL

> I am using the JSTL EL Validator that I wrote, but not the BSF or OGNL
> ones

I assume this plan involves running Rhino as the server side Javascript
engine?  I don't think that's a good idea.  

**Why? Size? Performance? Overhead? What if Rhino was only an option (as I
am suggesting)? Would you still be against Rhino if it is only an option not
a requirement?

Validator shouldn't have to
start up a Javascript engine on the server just to do some validations. 

** I agree. This is why I never suggested such a thing.
** It would not have to start JavaScript. This would be an *option*, it
would only start up JavaScript if you used the validate-javascript rule
(more likely validate-bsf).

Javascript is a client side validation tool that's a convenience to users.

** I agree.

 The real benefit from Validator is the server side checks it performs
which should remain in Java.

** I agree. This is an augmentation not a replacement. I am not suggesting
the that validator framework have any dependence on JavaScript/Rhino. This
would all be included and encapsulated in one validator rule.

** Hmmmm... what if you could do a simple expression like this....

form.passwordCheck != value 

Or something like this

form.startDate < value

and replace one ActionForm class (40+ lines of code). 

** Less code to write means less code to maintain.

** I think JSTL EL and JavaScript are valid options for doing comparison of
form fields. Using the BSF, the user could pick any Java scripting language
(Bean Shell, Jython, JRuby or whatever). Duplication is evil. It would be
nice to avoid duplication if possible.

** What were your thoughts on JSTL EL as an expression language for doing
form field relationship comparisons?

** I'll come up with a proof of concept for BSF later. For now I would love
to get your thoughts on JSTL EL.

David

<snip>



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to