I'm with you on the idea that tags should always be cached and should simply
clear their state if desired. Either that, or they should never be cached.
We could even pass the state Map in to the execute method along with the
Context... or something like that. I'd say that the next step is to put
together the list of proposed API changes and hash out where this is going
next.

-----Original Message-----
From: Paul Libbrecht [mailto:[EMAIL PROTECTED] 
Sent: Tuesday, July 26, 2005 12:11 PM
To: Jakarta Commons Developers List
Subject: Re: [jelly] 1.0-final problem wrt tag caching


Le 24 juil. 05, à 23:33, Hans Gilde a écrit :

> Got it, thanks. I was the one who accidentally removed isCacheTags(), 
> sorry.

By no means, I was the one and that is not so accidental.
(I remember somewhat warning about it)

Allow me to comment on the following, I think there's a deep issue here!
> Previous behavior was for each call to TagScript.run() to use a new 
> tag instance.

That is precisely a problem in several cases and tags should be "free 
to choose this".

- I believe Swing's ActionTag cannot address its containing component 
if the tag is gone away.

- I believe it is natural for the states of an http-client tag to be 
stored in the tag and not need a variable for this (e.g. for cookies)

- I remember it was a problem in define-tags but cannot nail this down 
correctly now.

I still believe that tags should, most of the time, be cached and would 
rather encourage end-do-tags-like clear of beans for things where it 
makes sense.

What does ant do ? I think it also caches them.

paul

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


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

Reply via email to