Re: [Webwork-devel] Webwork 1.0.3 / CVS taglib totally broken?

[Mike Cannon-B. <mi...@at...>]

Ok - I thought that fixed it but that is just the tip of the iceberg.

Pretty much every tag now 'resets' itself in the release() method - argh!

I'll try to prepare a fill set of diffs.

-mike

On 23/5/02 4:35 PM, "Mike Cannon-Brookes" (mi...@at...) penned the
words:

> Ok - I have found the exact commit that fucked everything up.
> 
> http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/webwork/webwork/src/main/webw
> ork/view/taglib/ui/ComponentTag.java.diff?r1=1.1&r2=1.2
> 
> BAD BAD BAD BAD BAD :)
> (apologies - just venting after spending 4 hours tracking this bastard down
> ;))
> 
> In my local copy of CVS I've removed the release() method of the
> ComponentTag, and put back the finally clause to the end of the doEndTag()
> method like so:
> 
> finally
> {
>  getStack().popValue();
>  params = new HashMap();
> }
> 
> This appears to make everything work again (I haven't had a chance to fully
> test it - but my test case now works).
> 
> Can we commit this to CVS and put out 1.0.4?
> 
> Cheers,
> Mike
> 
> PS I can't believe this hasn't been caught in months. Am we the only WW
> users using the UI tags with Orion and Resin? :) This would have broken
> pretty much all uses of the UI tags!
> 
> 
> Mike Cannon-Brookes
> mi...@at...
> 
> ATLASSIAN - Your J2EE Expert Partner
> --------------------------------------------------------
>> Brilliant Software - http://www.atlassian.com/software
>> Legendary Services - http://www.atlassian.com/support
> 
> On 23/5/02 4:23 PM, "Mike Cannon-Brookes" (mi...@at...) penned the
> words:
> 
>> I'm getting further in tracking this down.
>> 
>> It broke between 1.0.1 (which works fine) and 1.0.2 (which doesn't work).
>> 
>> I'll see if I can find the exact commits and create a patch for the current
>> CVS (to save rollbacks).
>> 
>> -mike
>> 
>> On 23/5/02 4:07 PM, "Mike Cannon-Brookes" (mi...@at...) penned the
>> words:
>> 
>>> I'm trying to what out what changes have occurred recently, but the taglib
>>> in CVS is currently broken and not backward compatible.
>>> 
>>> I think it's to do with the recent release() movements etc.
>>> 
>>> Values in correct containers (tested in both Orion and Resin) no longer get
>>> reset! 
>>> 
>>> Ie
>>> 
>>> textfield: text = "foo"
>>> ... Jsp ...
>>> textfield: text = "foo"
>>> (this should be "" as the value in the stack is null, but the tag still has
>>> the old value because it is no longer cleared)
>>> 
>>> I think this manifests itself in a number of places.
>>> 
>>> I'm betting (investigating the exact cause now) that this is because you're
>>> relying on release() being called everytime the tag is executed.
>>> 
>>> AFAIK (from memory I submitted a fix for this a while ago and the same error
>>> has crept in again) this is NOT what the specification specifies (I haven't
>>> got it with me) - this is how poor containers (Tomcrap) implement it.
>>> 
>>> release() is called by the container when it pleases, to release resources
>>> the tag is holding (the only time it must be called is before a GC I seem to
>>> recall). 
>>> 
>>> If you want to _reset_ values you should do this at the end of doEndTag().
>>> 
>>> I'll try to fix this and submit patches but I suggest a rollback to previous
>>> source (as it would be easier!). Any idea why this was changed? It worked
>>> perfectly for me before.
>>> 
>>> Maybe I should have just paid more attention on the devel list, or maybe I'm
>>> just v. tired ;)
>>> 
>>> -mike
>>> 
>>> 
>>> Mike Cannon-Brookes
>>> mi...@at...
>>> 
>>> ATLASSIAN - Your J2EE Expert Partner
>>> --------------------------------------------------------
>>>> Brilliant Software - http://www.atlassian.com/software
>>>> Legendary Services - http://www.atlassian.com/support
>>> 
>>> 
>>> _______________________________________________________________
>>> 
>>> Don't miss the 2002 Sprint PCS Application Developer's Conference
>>> August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm
>>> 
>>> _______________________________________________
>>> Webwork-devel mailing list
>>> Web...@li...
>>> https://lists.sourceforge.net/lists/listinfo/webwork-devel
>> 
>> 
>> _______________________________________________________________
>> 
>> Don't miss the 2002 Sprint PCS Application Developer's Conference
>> August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm
>> 
>> _______________________________________________
>> Webwork-devel mailing list
>> Web...@li...
>> https://lists.sourceforge.net/lists/listinfo/webwork-devel
> 
> 
> _______________________________________________________________
> 
> Don't miss the 2002 Sprint PCS Application Developer's Conference
> August 25-28 in Las Vegas -- http://devcon.sprintpcs.com/adp/index.cfm
> 
> _______________________________________________
> Webwork-devel mailing list
> Web...@li...
> https://lists.sourceforge.net/lists/listinfo/webwork-devel