Hello David !

Thanks for your answer.

If I have well understood, having a big web application ( responding to
multi customer devices ) , using STRUTS with modified intrisic functions
(in example those involved in solving device forward page choice), would
have no interest because of that :

Ressources needed for running a big multi devices application ( considering
a big struts-config.xml including all devices specifics
Views and Forms forwards ), would be more huge than those for the sames
devices specific application.
Assuming the application selection is made at fisrt by the device
corresponding gateway.


In other words :

Why to declare, in a big struts-config.xml, all the devices specific actions
( those forwarding to Forms, Views, etc...) when we could do
that in the respective and smallest separated applications's
struts-config.xml files ?

By the same way, in separated applications, the need to modify the struts
intrisic functions ( for example "forward",
to force the function to return the device normalized name page ) becomes a
non sense because we are yet in a device
specific struts-config.xml actions for views and forms.
The "forward" function, in this case,  don't have to return any other page
than the one we have indicated it !

STRUTS IS NOT HTML DEVICE SPECIFIC !  Only those for html are, but they are
usualy involved in Forms or Views,
precisely where the application stuff becomes device specific !
Logic and Action tags are html independent !

If I am right ( or instead, if I have understood a little bit more about
struts ), the Method 2 is the only and natural way, and I have
been exploring ( with Method 1 ) a very complicated and limitative way to
make struts have a device notion !

The way to manage 3 different customer devices, in an application is to make
3 different applications !

If it's true, it was so simple than I have to appologize because of my
inexperience of struts.

Thank you all, and specialy David, for confirming me that !


Manuel Vilar,






Manuel Vilar
Service developpement
01 45 15 03 32

-----Message d'origine-----
De : David M. Karr [mailto:[EMAIL PROTECTED]]
Envoyé : lundi 8 avril 2002 19:01
À : [EMAIL PROTECTED]
Objet : Re: TR: Multi Device application with STRUTS

>>>>> "Manuel" == Manuel Vilar <[EMAIL PROTECTED]> writes:

    Manuel> Hello,
    Manuel> First : I am a French developper, so I apologize for my poor
English !

    Manuel> Our society decided to use STRUTS because of it's MVC approach !

    Manuel> Our applications have to be accessed by differents customers
devices.
    Manuel> Each customer device supports different languages ( html, vdxml,
WML,
    Manuel> etc...)

    Manuel> What's on ?

    Manuel> STRUTS having not intrisinc customer device notion, we are
thinking about
    Manuel> two ways
    Manuel> for making STRUTS have a customer device notion.

I've often wondered (but not enough to write a RFE for it :) ) whether it
would
be useful to build an extension to the "forward" element, such that the
"forward" still has a single name that is referenced from code, but inside
the
element itself, it has a "switch" on a header, parameter, or something else,
which causes it to choose one of several real alternatives.

This would be less useful if you don't have close to a 1-1 mapping between
sets
of alternatives.  For example, this may not be very useful when the choices
are
between HTML and WML, as it's possible that a single HTML page would map to
more than one WML page (although you might have several WML cards in the
deck).

My aim in considering an approach like this is to avoid building
application-specific infrastructure to "make the choice", by soft-coding the
choices into the application configuration.

--
===================================================================
David M. Karr          ; Java/J2EE/XML/Unix/C++
[EMAIL PROTECTED]


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


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

Reply via email to