webwork-devel — WebWork development

Re: [Webwork-devel] developers?

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

Hehe *sees a bunch of American's dancing around yelling Jeri! Jeri! Jeri!*

Seriously Maurice - if you added this feature I would almost fly over there
and give you a huge hug. Or if that's too scary, perhaps some beer.

We've been looking at how to generate reports and I just found Jasper
yesterday. (The name is very confusing though because of the JSP compilation
module called Jasper - IMHO they should change it)

Integrating Jasper and WW - +10!

(In fact there aren't too many WW features I'd prefer more than this at the
moment)

-mike

PS Matt - JIRA is running on Resin and Orion sweetly now - which means I
think all the taglib problems are fixed! *fingers crossed*

 On 6/6/02 11:40 AM, "Matt Baldree" (ma...@sm...) penned the words:

> Why don't you go ahead and add this to WW's JERI so we have it on record.
> 
> http://opensource.atlassian.com/projects/webwork
> 
> ----- Original Message -----
> From: "Maurice Parker" <Ma...@Vi...>
> To: <ma...@sm...>
> Cc: <web...@li...>
> Sent: Wednesday, June 05, 2002 8:30 PM
> Subject: Re: [Webwork-devel] developers?
> 
> 
>> Hey!
>> 
>> Don't drop me, I'm not done yet.  :-)  I have been extremely busy with my
>> day job lately.  It has been one project deadline after another and I
>> haven't had time to even read the webwork-users list for months.
>> 
>> Look for me to do something sometime in the future.  For example I would
>> like to enable JasperReports to be used as a view technology so that PDF
>> report generation could be done using the same actions that you using to
>> generate HTML reports.
>> 
>> -Maurice
>> 
>> 
>> On Wednesday, June 5, 2002, at 07:42  AM, ma...@sm... wrote:
>> 
>>> Some of you have been pretty silent these days. Not
>>> sure if you are busy, moved on, or somethings else. So,
>>> in an effort to keep the developer list current, I
>>> would like to ask if you are no longer interested in
>>> developing for WW and/or don't see yourself developing
>>> for WW and would like your name removed as a developer
>>> then just drop me a reply and I'll take care of it. If
>>> you want to stay on, great.
>>> 
>>> This isn't an attempt to remove anybody but just spring
>>> cleaning :).
>>> 
>>> -Matt
>>> 
>>> 
>>> 
>>> _______________________________________________________________
>>> 
>>> 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

Re: [Webwork-devel] developers?

[Matt B. <ma...@sm...>]

Why don't you go ahead and add this to WW's JERI so we have it on record.

http://opensource.atlassian.com/projects/webwork

----- Original Message -----
From: "Maurice Parker" <Ma...@Vi...>
To: <ma...@sm...>
Cc: <web...@li...>
Sent: Wednesday, June 05, 2002 8:30 PM
Subject: Re: [Webwork-devel] developers?


> Hey!
>
> Don't drop me, I'm not done yet.  :-)  I have been extremely busy with my
> day job lately.  It has been one project deadline after another and I
> haven't had time to even read the webwork-users list for months.
>
> Look for me to do something sometime in the future.  For example I would
> like to enable JasperReports to be used as a view technology so that PDF
> report generation could be done using the same actions that you using to
> generate HTML reports.
>
> -Maurice
>
>
> On Wednesday, June 5, 2002, at 07:42  AM, ma...@sm... wrote:
>
> > Some of you have been pretty silent these days. Not
> > sure if you are busy, moved on, or somethings else. So,
> > in an effort to keep the developer list current, I
> > would like to ask if you are no longer interested in
> > developing for WW and/or don't see yourself developing
> > for WW and would like your name removed as a developer
> > then just drop me a reply and I'll take care of it. If
> > you want to stay on, great.
> >
> > This isn't an attempt to remove anybody but just spring
> > cleaning :).
> >
> > -Matt
> >
> >
> >
> > _______________________________________________________________
> >
> > 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
> >
>
>
>

Re: [Webwork-devel] developers?

[Matt B. <ma...@sm...>]

Damn! I wonder where you've been :).

----- Original Message -----
From: "Maurice Parker" <Ma...@Vi...>
To: <ma...@sm...>
Cc: <web...@li...>
Sent: Wednesday, June 05, 2002 8:30 PM
Subject: Re: [Webwork-devel] developers?


> Hey!
>
> Don't drop me, I'm not done yet.  :-)  I have been extremely busy with my
> day job lately.  It has been one project deadline after another and I
> haven't had time to even read the webwork-users list for months.
>
> Look for me to do something sometime in the future.  For example I would
> like to enable JasperReports to be used as a view technology so that PDF
> report generation could be done using the same actions that you using to
> generate HTML reports.
>
> -Maurice
>
>
> On Wednesday, June 5, 2002, at 07:42  AM, ma...@sm... wrote:
>
> > Some of you have been pretty silent these days. Not
> > sure if you are busy, moved on, or somethings else. So,
> > in an effort to keep the developer list current, I
> > would like to ask if you are no longer interested in
> > developing for WW and/or don't see yourself developing
> > for WW and would like your name removed as a developer
> > then just drop me a reply and I'll take care of it. If
> > you want to stay on, great.
> >
> > This isn't an attempt to remove anybody but just spring
> > cleaning :).
> >
> > -Matt
> >
> >
> >
> > _______________________________________________________________
> >
> > 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
> >
>
>
>

Re: [Webwork-devel] developers?

[Maurice P. <Ma...@Vi...>]

Hey!

Don't drop me, I'm not done yet.  :-)  I have been extremely busy with my 
day job lately.  It has been one project deadline after another and I 
haven't had time to even read the webwork-users list for months.

Look for me to do something sometime in the future.  For example I would 
like to enable JasperReports to be used as a view technology so that PDF 
report generation could be done using the same actions that you using to 
generate HTML reports.

-Maurice


On Wednesday, June 5, 2002, at 07:42  AM, ma...@sm... wrote:

> Some of you have been pretty silent these days. Not
> sure if you are busy, moved on, or somethings else. So,
> in an effort to keep the developer list current, I
> would like to ask if you are no longer interested in
> developing for WW and/or don't see yourself developing
> for WW and would like your name removed as a developer
> then just drop me a reply and I'll take care of it. If
> you want to stay on, great.
>
> This isn't an attempt to remove anybody but just spring
> cleaning :).
>
> -Matt
>
>
>
> _______________________________________________________________
>
> 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
>

Re: [Webwork-devel] developers?

[<ma...@sm...>]

Clarification -

I posted this to the developer list in general but the
e-mail is more specifically focused for those that have
CVS access. I could have sent it directly to them but I
was lazy :).

-Matt

On Wed, 05 June 2002, ma...@sm... wrote

> 
> Some of you have been pretty silent these days. Not
> sure if you are busy, moved on, or somethings else.
So,
> in an effort to keep the developer list current, I
> would like to ask if you are no longer interested in
> developing for WW and/or don't see yourself developing
> for WW and would like your name removed as a developer
> then just drop me a reply and I'll take care of it. If
> you want to stay on, great.
> 
> This isn't an attempt to remove anybody but just
spring
> cleaning :).
> 
> -Matt
> 
> 
> 
>
_______________________________________________________________
> 
> 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

[Webwork-devel] developers?

[<ma...@sm...>]

Some of you have been pretty silent these days. Not
sure if you are busy, moved on, or somethings else. So,
in an effort to keep the developer list current, I
would like to ask if you are no longer interested in
developing for WW and/or don't see yourself developing
for WW and would like your name removed as a developer
then just drop me a reply and I'll take care of it. If
you want to stay on, great.

This isn't an attempt to remove anybody but just spring
cleaning :).

-Matt

[Webwork-devel] JIRA

[<ma...@sm...>]

I populated JIRA with our TODO list.

Please take a moment to visit the project page and grab
a user account. You should be able to browse the
project without an account, but if you want to add
issues, you will need an account. It will take a few
turns with this before we are up and going, but I think
it will be well worth the effort.

-Matt

http://opensource.atlassian.com/projects/webwork

Re: [Webwork-devel] ActionSupport.command

[Matt B. <ma...@sm...>]

done

----- Original Message ----- 
From: <wa...@mw...>
To: <web...@li...>
Sent: Tuesday, May 28, 2002 3:40 PM
Subject: [Webwork-devel] ActionSupport.command


> Would it be possible to change the command String in ActionSupport to 
> protected from private.  
> 
> It would make it easier to subclass ActionSupport.
> 
> 
> Thanks,
> 
> Walt
> *****************************************************
> Walter Szewelanczyk
> IS Director
> M.W. Sewall & CO.        email : wa...@mw...     
> 259 Front St.            Phone : (207) 442-7994 x 128
> Bath, ME 04530           Fax   : (207) 443-6284
> *****************************************************
> 
> 
> _______________________________________________________________
> 
> 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
> 
> 

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

[Matt B. <ma...@sm...>]

Ok. I overhauled taglib to remove release() except where necessary, which
was only a few places. I also added a work around for Resin 2.1.1 bug in the
ValueStack. I tested against Tomcat 4.0.4b2, WebLogic 7.0, and Resin 2.1.1.
I also added a FAQ for installing WW's WAR in Resin 2.1.1. I suggest you
read this if you are using Resin.

Please, take a second to check the latest CVS against your container to make
sure everything works fine. Sorry about the delay, it's been a busy week.

-Matt

----- Original Message -----
From: "Mike Cannon-Brookes" <mi...@at...>
To: <web...@li...>
Sent: Monday, May 27, 2002 9:13 PM
Subject: Re: [Webwork-devel] Re: Webwork 1.0.3 / CVS taglib totally broken?


> Attached is an AbstractValueStack that fixes one of the NPEs.
>
> I looked into the ValueStack one for ages and I ended up working around it
> in my code instead of fixing it - but here's my analysis so far:
> - the actual NPE is thrown from inside resin's QPageContext class
> - calling context.getRequest() just before context.findAttribute(x)
returns
> a null, so the problem is obviously that at this point the context's
request
> object is null
> - this only happens when I try to refer to an action from within a
component
> tag (so it may be the include process loses the request somehow?), let me
> explain in pseudo JSP
>
> <ww:action id="foo" />
> <ww:select ... list="@foo/list" />
>
> - the call to @foo/list inside the value stack here fails with the NPE
> above.
>
> Hope this helps someone else debug it! For now I've removed the reliance
on
> actions of that piece of code (it now looks in the page action and works
> fine)
>
> Hope this helps!
>
> Cheers,
> Mike
>
> PS I also had to set valueAttr to null at the end of PropertyTag to get it
> to work properly
>
> On 28/5/02 4:21 AM, "Matt Baldree" (ma...@sm...) penned the words:
>
> > I've checked in a Resin 2.1.1 test case that highlights the NPE problem.
If
> > you click on the test it will work but if you uncomment line 81 in
> > tests/test-resin.jsp you will get the NPE. I can't see the problem right
now
> > so I committed this test case incase anybody else felt motivated to give
it
> > a look. The test runs fine on other containers so I'm not sure what
Resin's
> > problem is.
> >
> > -Matt
> >
> > ----- Original Message -----
> > From: "Matt Baldree" <ma...@sm...>
> > To: "Sven Kuenzler" <sv...@im...>
> > Cc: "Webwork-Developer" <web...@li...>
> > Sent: Monday, May 27, 2002 12:41 PM
> > Subject: Fw: [Webwork-devel] Re: Webwork 1.0.3 / CVS taglib totally
broken?
> >
> >
> >> It is happening in both places for Resin 2.1.1. so I don't believe it
is a
> >> VS problem. I believe it is a tag problem. I'm looking into it.
> >>
> >> ----- Original Message -----
> >> From: "Sven Kuenzler" <sv...@im...>
> >> Newsgroups: gmane.comp.java.webwork.devel
> >> To: <web...@li...>
> >> Sent: Monday, May 27, 2002 12:13 PM
> >> Subject: [Webwork-devel] Re: Webwork 1.0.3 / CVS taglib totally broken?
> >>
> >>
> >>> "Mike Cannon-Brookes" <mi...@at...> schrieb im Newsbeitrag
> >>> news:B9182255.B7E7%mi...@at......
> >>>> PS More investigating (this time in Resin) since I wrote the above -
> >> there
> >>>> are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
> >>>> ValueStack.java and line 403 of AbstractValueStack.java (both
probably
> >>>> related to null values being looked up and not checked for before
> >> calling
> >>>> method on the value object)
> >>>
> >>> Strange. Both lines you mention are unchanged to the old ValueStack
> >> Besides
> >>> the fact that some of them now live in an abstract class, that is. Are
> > you
> >>> sure the NPE's aren't occurring with the old ValueStack as well?
> >>>
> >>>     Sven....
> >>>
> >>>
> >>>
> >>>
> >>>
> >>> _______________________________________________________________
> >>>
> >>> 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
>
>

Re: [Webwork-devel] Taglib guidelines

[Matt B. <ma...@sm...>]

Thanks for the link. I'll try and bring these up to suggested guidelines by
week's end.

----- Original Message -----
From: "Mike Cannon-Brookes" <mi...@at...>
To: <web...@li...>
Sent: Wednesday, May 29, 2002 8:25 PM
Subject: [Webwork-devel] Taglib guidelines


> Hey guys,
>
> Just found these (care of Shawn Bayern) and thought they would be of use
to
> WW. They're from the Jakarta taglibs project, which like it or hate it,
has
> a lot of experience in developing tags.
>
> http://jakarta.apache.org/taglibs/guidelines.html
>
> I believe the WW tags violate at least one of them (my release()
troubles!)
> but it's a great overall summary of things to watch out for.
>
> Enjoy.
>
> Cheers,
> Mike
>
> ATLASSIAN - Expert J2EE Software, Services and Support
> ------------------------------------------------------
> http://www.atlassian.com
>
> JIRA is the simple, powerful way to track and manage issues
> - http://www.atlassian.com/software/jira
>
>
> _______________________________________________________________
>
> 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
>
>

[Webwork-devel] Taglib guidelines

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

Hey guys,

Just found these (care of Shawn Bayern) and thought they would be of use to
WW. They're from the Jakarta taglibs project, which like it or hate it, has
a lot of experience in developing tags.

http://jakarta.apache.org/taglibs/guidelines.html

I believe the WW tags violate at least one of them (my release() troubles!)
but it's a great overall summary of things to watch out for.

Enjoy.

Cheers,
Mike

ATLASSIAN - Expert J2EE Software, Services and Support
------------------------------------------------------
http://www.atlassian.com

JIRA is the simple, powerful way to track and manage issues
- http://www.atlassian.com/software/jira

[Webwork-devel] Cocoon2 integration

[Sven K. <sv...@im...>]

I have now put together all things needed so you can have a look the C2
stuff I've done.
You can download the bundle from http://www.tzi.de/~svenk/webcoon.zip

Unzip the package, do a "build jar", deploy work\jar\webcoon.war in your
favourite servlet container and surf to http://host/webcoon
This was successfully tested with Tomcat 4.0.4-b2 standalone It  failed on
Jboss-Catalina 3.0 RC. JBoss+Catalina 2.4.4 works.

If you'd like to take a look behind the scenes, everything you need is in
resources\web\mount\webwork. Esp. sitemap.xmap is of interest.

Currently, the Cocoon folks are discussing a new InputModule concept which
might be useful for my work. Because of this and due to personal time
constraints I have stopped work on this for now.

So far, I have implemented as much of the ui-tags as needed to get the bank
app running. Notable changes are
- the themability of ui tags will be implemented using xslt. A first shot at
this is included in the bank app for example.
- webwork:include support is completely missing ATM. I am not entirely sure
how to do this yet. Maybe one could use C2's XInclude transformer or the
sitemap's aggregation concept.

Comments, suggestions and improvements welcome :-)

    Sven....

[Webwork-devel] ActionSupport.command

[<wa...@mw...>]

Would it be possible to change the command String in ActionSupport to 
protected from private.  

It would make it easier to subclass ActionSupport.


Thanks,

Walt
*****************************************************
Walter Szewelanczyk
IS Director
M.W. Sewall & CO.        email : wa...@mw...     
259 Front St.            Phone : (207) 442-7994 x 128
Bath, ME 04530           Fax   : (207) 443-6284
*****************************************************

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

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

Attached is an AbstractValueStack that fixes one of the NPEs.

I looked into the ValueStack one for ages and I ended up working around it
in my code instead of fixing it - but here's my analysis so far:
- the actual NPE is thrown from inside resin's QPageContext class
- calling context.getRequest() just before context.findAttribute(x) returns
a null, so the problem is obviously that at this point the context's request
object is null
- this only happens when I try to refer to an action from within a component
tag (so it may be the include process loses the request somehow?), let me
explain in pseudo JSP

<ww:action id="foo" />
<ww:select ... list="@foo/list" />

- the call to @foo/list inside the value stack here fails with the NPE
above.

Hope this helps someone else debug it! For now I've removed the reliance on
actions of that piece of code (it now looks in the page action and works
fine)

Hope this helps!

Cheers,
Mike

PS I also had to set valueAttr to null at the end of PropertyTag to get it
to work properly

On 28/5/02 4:21 AM, "Matt Baldree" (ma...@sm...) penned the words:

> I've checked in a Resin 2.1.1 test case that highlights the NPE problem. If
> you click on the test it will work but if you uncomment line 81 in
> tests/test-resin.jsp you will get the NPE. I can't see the problem right now
> so I committed this test case incase anybody else felt motivated to give it
> a look. The test runs fine on other containers so I'm not sure what Resin's
> problem is.
> 
> -Matt
> 
> ----- Original Message -----
> From: "Matt Baldree" <ma...@sm...>
> To: "Sven Kuenzler" <sv...@im...>
> Cc: "Webwork-Developer" <web...@li...>
> Sent: Monday, May 27, 2002 12:41 PM
> Subject: Fw: [Webwork-devel] Re: Webwork 1.0.3 / CVS taglib totally broken?
> 
> 
>> It is happening in both places for Resin 2.1.1. so I don't believe it is a
>> VS problem. I believe it is a tag problem. I'm looking into it.
>> 
>> ----- Original Message -----
>> From: "Sven Kuenzler" <sv...@im...>
>> Newsgroups: gmane.comp.java.webwork.devel
>> To: <web...@li...>
>> Sent: Monday, May 27, 2002 12:13 PM
>> Subject: [Webwork-devel] Re: Webwork 1.0.3 / CVS taglib totally broken?
>> 
>> 
>>> "Mike Cannon-Brookes" <mi...@at...> schrieb im Newsbeitrag
>>> news:B9182255.B7E7%mi...@at......
>>>> PS More investigating (this time in Resin) since I wrote the above -
>> there
>>>> are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
>>>> ValueStack.java and line 403 of AbstractValueStack.java (both probably
>>>> related to null values being looked up and not checked for before
>> calling
>>>> method on the value object)
>>> 
>>> Strange. Both lines you mention are unchanged to the old ValueStack
>> Besides
>>> the fact that some of them now live in an abstract class, that is. Are
> you
>>> sure the NPE's aren't occurring with the old ValueStack as well?
>>> 
>>>     Sven....
>>> 
>>> 
>>> 
>>> 
>>> 
>>> _______________________________________________________________
>>> 
>>> 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

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

[Matt B. <ma...@sm...>]

Give me a few days to look at this. I'm not oppose to it. I just need time
to test it all.

-Matt

----- Original Message -----
From: "Mike Cannon-Brookes" <mi...@at...>
To: <web...@li...>
Sent: Monday, May 27, 2002 7:26 PM
Subject: Re: [Webwork-devel] Webwork 1.0.3 / CVS taglib totally broken?


> On 27/5/02 9:38 PM, "Matt Baldree" (ma...@sm...) penned the words:
>
> > I agree that reset logic should be moved to doEndTag() but this is
different
> > from null'ing attributes. Null'ing attributes should still be in
release().
> > I don't want to make the blanket change at this time. I did fix Property
and
> > Action. I would prefer to make a couple of fixes and all containers
should
> > work.
>
> I don't understand. Every webwork tag (that I can see) assumes that
> attributes are null at the start of doStartTag(). Moving everything into
> reset() and calling from doEndTag() and release() fixes this problem for
all
> containers!
>
> Why would you not want to make this change? :)
>
> > I changed the escaping of outer quotes to true by default. The reason
> > it was false was just a first guess. I guessed wrong :). I haven't
tested
> > these changes on Resin yet. I can't get to caucho but I will by early
this
> > week.
>
> Cool - I'll update my CVS and try again today - another day battling
Resin.
> ;)
>
> Cheers,
> Mike
>
> > -Matt
> >
> > ----- Original Message -----
> > From: "Mike Cannon-Brookes" <mi...@at...>
> > To: <ma...@sm...>
> > Cc: <web...@li...>
> > Sent: Monday, May 27, 2002 2:50 AM
> > Subject: Re: [Webwork-devel] Webwork 1.0.3 / CVS taglib totally broken?
> >
> >
> >> Ok - I've checked and from preliminary looks it seems like the
> > ComponentTag
> >> is fixed in Orion (haven't tested Resin yet - will do)
> >>
> >> PropertyTag:
> >> I had to fix the PropertyTag again though.
> >>
> >> Simple fix: (can't check in myself but here's the details)
> >> - remove return statement that's not at the end of the doEndTag method)
> >> - call release() manually before return EVAL_PAGE;
> >> (see attachment)
> >>
> >> Again - the problem here is that data is reset in the release() method
> > which
> >> is incorrect ;)
> >>
> >> I have a suggested fix for every tag - rename ALL the release() methods
to
> >> reset() and call them manually at the end of the doEndTag() methods (or
> >> before every return EVAL_PAGE statement) as well. This keeps the
'reset'
> >> logic compartmentalised but at the same time makes sure that it will be
> >> called on ALL containers.
> >>
> >> I've done this in the attached file so you can see what I mean.
> >>
> >> TextUtil:
> >> Also I changed the default value of "escapeOuterQuotes" to true. Why
would
> >> you ever want this to be false? (Having it false by default means for
> >> example that any formfield with "text" as a value only gets text which
is
> > v.
> >> dangerous) Can we change this in CVS? Why was escapeOuterQuotes ever
> > needed?
> >>
> >>
> >> I'll yell again if I find any more specific problems - but I can see
that
> >> the release() method does things in other tag (which I probably don't
use)
> >> which is bad.
> >>
> >> Cheers,
> >> Mike
> >>
> >> PS More investigating (this time in Resin) since I wrote the above -
there
> >> are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
> >> ValueStack.java and line 403 of AbstractValueStack.java (both probably
> >> related to null values being looked up and not checked for before
calling
> >> method on the value object)
> >>
> >>
> >> 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 24/5/02 1:03 AM, "ma...@sm..." (ma...@sm...) penned
the
> >> words:
> >>
> >>> Alright. Thanks for the reply. We'll get this worked
> >>> out and commented accordingly. Whatever the outcome,
> >>> I'll check it against Resin, WL, and Tomcat. I'll leave
> >>> you and others to check it against Orion. As soon as I
> >>> get a thumbs up from the community, I'll release a
> >>> patch 1.0.4.
> >>>
> >>> -Matt
> >>>
> >>> On Fri, 24 May 2002, Mike Cannon-Brookes wrote
> >>>
> >>>>
> >>>> Matt,
> >>>>
> >>>> Thanks mate - I'll check the CVS in the morning and
> >>> report back. I found a
> >>>> bunch of changes I needed to make today to get it
> >>> working, so I'll see how
> >>>> CVS works for me and start from there again tomorrow
> >>> (with more mental
> >>>> energy).
> >>>>
> >>>> Re: release() - this is NOT the preferred method at
> >>> all ;)
> >>>>
> >>>> Again (yes, I've checked now with the spec and the
> >>> author of JSP Tags book)
> >>>> release() is called by the container to release
> >>> resources associated with
> >>>> the tag. It can be called at ANY time the container
> >>> wishes (you have no
> >>>> guarantee of when it is called). The only restriction
> >>> is that it must be
> >>>> called before the tag instance is garbage collected.
> >>>>
> >>>> So - what you have is lazy programmers (WL, Tomcat
> >>> are the two glaring
> >>>> examples) who just call release() after every tag
> >>> invocation - which leads
> >>>> to people's misplaced belief that it MUST be called
> >>> everytime. This is not
> >>>> so! :) Fast containers (like Resin and Orion) pool
> >>> tag instances and reuse
> >>>> them (maximising speed), and call release() as few
> >>> times as possible
> >>>> (sometimes never). I'm sure it's harder to code it
> >>> this way, but the end
> >>>> result is speed.
> >>>>
> >>>> Net result:
> >>>>
> >>>>    NEVER do anything you want to happen in release()!
> >>>>
> >>>> This is true across all the JSP tags in WW and other
> >>> projects. If you want
> >>>> logic to occur, or variables to be reset for each tag
> >>> invocation, reset them
> >>>> at the end of the doEndTag() method (either before
> >>> returning, or as a
> >>>> finally block).
> >>>>
> >>>> I'll check CVS in the morning - good work on the
> >>> quick fixes though!
> >>>>
> >>>> Cheers,
> >>>> Mike
> >>>>
> >>>> On 23/5/02 10:20 PM, "ma...@sm..."
> >>> (ma...@sm...) penned the
> >>>> words:
> >>>>
> >>>>> For WebLogic, using release is the preferred
> >>> method. I
> >>>>> disagree with your blanket statement. I fixed the
> >>>>> release order in the latest CVS and in the branch.
> >>> I'm
> >>>>> told that it is working correctly on Orion but I'm
> >>>>> waiting for official confirmation. I would
> >>> appreciate
> >>>>> if you could check it with the latest CVS and let me
> >>>>> know.
> >>>>>
> >>>>> On Thu, 23 May 2002, Mike Cannon-Brookes wrote
> >>>>>
> >>>>>>
> >>>>>> Yes - release() is the bane of tag developers.
> >>>>>>
> >>>>>> The simple rule is basically - if you're using
> >>>>> release, you're doing things
> >>>>>> wrong!
> >>>>>>
> >>>>>> -mike
> >>>>>>
> >>>>>> On 23/5/02 4:28 PM, "Heng Sin Low"
> >>>>> (low...@ya...) penned the words:
> >>>>>>
> >>>>>>> Ha, finally u got what I mean :)
> >>>>>>>
> >>>>>>> Yes, doEndTag should be the right place. I
> >>> remember
> >>>>>>> Matt have done a massive check in recently that
> >>> move
> >>>>>>> all this resetting stuff to the release() method.
> >>>>>>>
> >>>>>>> Regards,
> >>>>>>> Low
> >>>>>>>
> >>>>>>>> I'm trying to what out what changes have occurred
> >>>>>>> recently, but the taglib
> >>>>>>> in CVS is currently broken and not backward
> >>>>>>> compatible.
> >>>>>>>
> >>>>>>>
> >>>>>>> __________________________________________________
> >>>>>>> Do You Yahoo!?
> >>>>>>> LAUNCH - Your Yahoo! Music Experience
> >>>>>>> http://launch.yahoo.com
> >>>>>>>
> >>>>>>>
> >>>>>
> >>> _______________________________________________________________
> >>>>>>>
> >>>>>>> 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
> >>>
> >>>
> >>>
> >>> _______________________________________________________________
> >>>
> >>> 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
>

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

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

I just looked at this again and the PropertyTag is _still_ broken I think.
Somehow I don't seem to be getting through here - my fix worked perfectly on
all containers, why was it not used?! :)

You still have valueAttr being set to null in the release() method.

However, property tag can be used like so <webwork:property /> but in this
case (as the valueAttr still has the value from the last time property was
used) the wrong value is used!

<ww:property value="foo">
    <ww:property value="../bar" />
    <ww:property /> <-- this should print use foo as the value correct? But
here it will use the 'cached' ../bar!
</ww:property>

Once again - everything should be reset in doEndTag! I cannot find a single
tag (browsing the ww source) that should EVER use release()!

-mike

On 27/5/02 9:38 PM, "Matt Baldree" (ma...@sm...) penned the words:

> I agree that reset logic should be moved to doEndTag() but this is different
> from null'ing attributes. Null'ing attributes should still be in release().
> I don't want to make the blanket change at this time. I did fix Property and
> Action. I would prefer to make a couple of fixes and all containers should
> work. I changed the escaping of outer quotes to true by default. The reason
> it was false was just a first guess. I guessed wrong :). I haven't tested
> these changes on Resin yet. I can't get to caucho but I will by early this
> week.
> 
> -Matt
> 
> ----- Original Message -----
> From: "Mike Cannon-Brookes" <mi...@at...>
> To: <ma...@sm...>
> Cc: <web...@li...>
> Sent: Monday, May 27, 2002 2:50 AM
> Subject: Re: [Webwork-devel] Webwork 1.0.3 / CVS taglib totally broken?
> 
> 
>> Ok - I've checked and from preliminary looks it seems like the
> ComponentTag
>> is fixed in Orion (haven't tested Resin yet - will do)
>> 
>> PropertyTag:
>> I had to fix the PropertyTag again though.
>> 
>> Simple fix: (can't check in myself but here's the details)
>> - remove return statement that's not at the end of the doEndTag method)
>> - call release() manually before return EVAL_PAGE;
>> (see attachment)
>> 
>> Again - the problem here is that data is reset in the release() method
> which
>> is incorrect ;)
>> 
>> I have a suggested fix for every tag - rename ALL the release() methods to
>> reset() and call them manually at the end of the doEndTag() methods (or
>> before every return EVAL_PAGE statement) as well. This keeps the 'reset'
>> logic compartmentalised but at the same time makes sure that it will be
>> called on ALL containers.
>> 
>> I've done this in the attached file so you can see what I mean.
>> 
>> TextUtil:
>> Also I changed the default value of "escapeOuterQuotes" to true. Why would
>> you ever want this to be false? (Having it false by default means for
>> example that any formfield with "text" as a value only gets text which is
> v.
>> dangerous) Can we change this in CVS? Why was escapeOuterQuotes ever
> needed?
>> 
>> 
>> I'll yell again if I find any more specific problems - but I can see that
>> the release() method does things in other tag (which I probably don't use)
>> which is bad.
>> 
>> Cheers,
>> Mike
>> 
>> PS More investigating (this time in Resin) since I wrote the above - there
>> are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
>> ValueStack.java and line 403 of AbstractValueStack.java (both probably
>> related to null values being looked up and not checked for before calling
>> method on the value object)
>> 
>> 
>> 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 24/5/02 1:03 AM, "ma...@sm..." (ma...@sm...) penned the
>> words:
>> 
>>> Alright. Thanks for the reply. We'll get this worked
>>> out and commented accordingly. Whatever the outcome,
>>> I'll check it against Resin, WL, and Tomcat. I'll leave
>>> you and others to check it against Orion. As soon as I
>>> get a thumbs up from the community, I'll release a
>>> patch 1.0.4.
>>> 
>>> -Matt
>>> 
>>> On Fri, 24 May 2002, Mike Cannon-Brookes wrote
>>> 
>>>> 
>>>> Matt,
>>>> 
>>>> Thanks mate - I'll check the CVS in the morning and
>>> report back. I found a
>>>> bunch of changes I needed to make today to get it
>>> working, so I'll see how
>>>> CVS works for me and start from there again tomorrow
>>> (with more mental
>>>> energy).
>>>> 
>>>> Re: release() - this is NOT the preferred method at
>>> all ;)
>>>> 
>>>> Again (yes, I've checked now with the spec and the
>>> author of JSP Tags book)
>>>> release() is called by the container to release
>>> resources associated with
>>>> the tag. It can be called at ANY time the container
>>> wishes (you have no
>>>> guarantee of when it is called). The only restriction
>>> is that it must be
>>>> called before the tag instance is garbage collected.
>>>> 
>>>> So - what you have is lazy programmers (WL, Tomcat
>>> are the two glaring
>>>> examples) who just call release() after every tag
>>> invocation - which leads
>>>> to people's misplaced belief that it MUST be called
>>> everytime. This is not
>>>> so! :) Fast containers (like Resin and Orion) pool
>>> tag instances and reuse
>>>> them (maximising speed), and call release() as few
>>> times as possible
>>>> (sometimes never). I'm sure it's harder to code it
>>> this way, but the end
>>>> result is speed.
>>>> 
>>>> Net result:
>>>> 
>>>>    NEVER do anything you want to happen in release()!
>>>> 
>>>> This is true across all the JSP tags in WW and other
>>> projects. If you want
>>>> logic to occur, or variables to be reset for each tag
>>> invocation, reset them
>>>> at the end of the doEndTag() method (either before
>>> returning, or as a
>>>> finally block).
>>>> 
>>>> I'll check CVS in the morning - good work on the
>>> quick fixes though!
>>>> 
>>>> Cheers,
>>>> Mike
>>>> 
>>>> On 23/5/02 10:20 PM, "ma...@sm..."
>>> (ma...@sm...) penned the
>>>> words:
>>>> 
>>>>> For WebLogic, using release is the preferred
>>> method. I
>>>>> disagree with your blanket statement. I fixed the
>>>>> release order in the latest CVS and in the branch.
>>> I'm
>>>>> told that it is working correctly on Orion but I'm
>>>>> waiting for official confirmation. I would
>>> appreciate
>>>>> if you could check it with the latest CVS and let me
>>>>> know.
>>>>> 
>>>>> On Thu, 23 May 2002, Mike Cannon-Brookes wrote
>>>>> 
>>>>>> 
>>>>>> Yes - release() is the bane of tag developers.
>>>>>> 
>>>>>> The simple rule is basically - if you're using
>>>>> release, you're doing things
>>>>>> wrong!
>>>>>> 
>>>>>> -mike
>>>>>> 
>>>>>> On 23/5/02 4:28 PM, "Heng Sin Low"
>>>>> (low...@ya...) penned the words:
>>>>>> 
>>>>>>> Ha, finally u got what I mean :)
>>>>>>> 
>>>>>>> Yes, doEndTag should be the right place. I
>>> remember
>>>>>>> Matt have done a massive check in recently that
>>> move
>>>>>>> all this resetting stuff to the release() method.
>>>>>>> 
>>>>>>> Regards,
>>>>>>> Low
>>>>>>> 
>>>>>>>> I'm trying to what out what changes have occurred
>>>>>>> recently, but the taglib
>>>>>>> in CVS is currently broken and not backward
>>>>>>> compatible.
>>>>>>> 
>>>>>>> 
>>>>>>> __________________________________________________
>>>>>>> Do You Yahoo!?
>>>>>>> LAUNCH - Your Yahoo! Music Experience
>>>>>>> http://launch.yahoo.com
>>>>>>> 
>>>>>>> 
>>>>> 
>>> _______________________________________________________________
>>>>>>> 
>>>>>>> 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
>>> 
>>> 
>>> 
>>> _______________________________________________________________
>>> 
>>> 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

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

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

On 27/5/02 9:38 PM, "Matt Baldree" (ma...@sm...) penned the words:

> I agree that reset logic should be moved to doEndTag() but this is different
> from null'ing attributes. Null'ing attributes should still be in release().
> I don't want to make the blanket change at this time. I did fix Property and
> Action. I would prefer to make a couple of fixes and all containers should
> work. 

I don't understand. Every webwork tag (that I can see) assumes that
attributes are null at the start of doStartTag(). Moving everything into
reset() and calling from doEndTag() and release() fixes this problem for all
containers!

Why would you not want to make this change? :)

> I changed the escaping of outer quotes to true by default. The reason
> it was false was just a first guess. I guessed wrong :). I haven't tested
> these changes on Resin yet. I can't get to caucho but I will by early this
> week.

Cool - I'll update my CVS and try again today - another day battling Resin.
;)

Cheers,
Mike

> -Matt
> 
> ----- Original Message -----
> From: "Mike Cannon-Brookes" <mi...@at...>
> To: <ma...@sm...>
> Cc: <web...@li...>
> Sent: Monday, May 27, 2002 2:50 AM
> Subject: Re: [Webwork-devel] Webwork 1.0.3 / CVS taglib totally broken?
> 
> 
>> Ok - I've checked and from preliminary looks it seems like the
> ComponentTag
>> is fixed in Orion (haven't tested Resin yet - will do)
>> 
>> PropertyTag:
>> I had to fix the PropertyTag again though.
>> 
>> Simple fix: (can't check in myself but here's the details)
>> - remove return statement that's not at the end of the doEndTag method)
>> - call release() manually before return EVAL_PAGE;
>> (see attachment)
>> 
>> Again - the problem here is that data is reset in the release() method
> which
>> is incorrect ;)
>> 
>> I have a suggested fix for every tag - rename ALL the release() methods to
>> reset() and call them manually at the end of the doEndTag() methods (or
>> before every return EVAL_PAGE statement) as well. This keeps the 'reset'
>> logic compartmentalised but at the same time makes sure that it will be
>> called on ALL containers.
>> 
>> I've done this in the attached file so you can see what I mean.
>> 
>> TextUtil:
>> Also I changed the default value of "escapeOuterQuotes" to true. Why would
>> you ever want this to be false? (Having it false by default means for
>> example that any formfield with "text" as a value only gets text which is
> v.
>> dangerous) Can we change this in CVS? Why was escapeOuterQuotes ever
> needed?
>> 
>> 
>> I'll yell again if I find any more specific problems - but I can see that
>> the release() method does things in other tag (which I probably don't use)
>> which is bad.
>> 
>> Cheers,
>> Mike
>> 
>> PS More investigating (this time in Resin) since I wrote the above - there
>> are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
>> ValueStack.java and line 403 of AbstractValueStack.java (both probably
>> related to null values being looked up and not checked for before calling
>> method on the value object)
>> 
>> 
>> 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 24/5/02 1:03 AM, "ma...@sm..." (ma...@sm...) penned the
>> words:
>> 
>>> Alright. Thanks for the reply. We'll get this worked
>>> out and commented accordingly. Whatever the outcome,
>>> I'll check it against Resin, WL, and Tomcat. I'll leave
>>> you and others to check it against Orion. As soon as I
>>> get a thumbs up from the community, I'll release a
>>> patch 1.0.4.
>>> 
>>> -Matt
>>> 
>>> On Fri, 24 May 2002, Mike Cannon-Brookes wrote
>>> 
>>>> 
>>>> Matt,
>>>> 
>>>> Thanks mate - I'll check the CVS in the morning and
>>> report back. I found a
>>>> bunch of changes I needed to make today to get it
>>> working, so I'll see how
>>>> CVS works for me and start from there again tomorrow
>>> (with more mental
>>>> energy).
>>>> 
>>>> Re: release() - this is NOT the preferred method at
>>> all ;)
>>>> 
>>>> Again (yes, I've checked now with the spec and the
>>> author of JSP Tags book)
>>>> release() is called by the container to release
>>> resources associated with
>>>> the tag. It can be called at ANY time the container
>>> wishes (you have no
>>>> guarantee of when it is called). The only restriction
>>> is that it must be
>>>> called before the tag instance is garbage collected.
>>>> 
>>>> So - what you have is lazy programmers (WL, Tomcat
>>> are the two glaring
>>>> examples) who just call release() after every tag
>>> invocation - which leads
>>>> to people's misplaced belief that it MUST be called
>>> everytime. This is not
>>>> so! :) Fast containers (like Resin and Orion) pool
>>> tag instances and reuse
>>>> them (maximising speed), and call release() as few
>>> times as possible
>>>> (sometimes never). I'm sure it's harder to code it
>>> this way, but the end
>>>> result is speed.
>>>> 
>>>> Net result:
>>>> 
>>>>    NEVER do anything you want to happen in release()!
>>>> 
>>>> This is true across all the JSP tags in WW and other
>>> projects. If you want
>>>> logic to occur, or variables to be reset for each tag
>>> invocation, reset them
>>>> at the end of the doEndTag() method (either before
>>> returning, or as a
>>>> finally block).
>>>> 
>>>> I'll check CVS in the morning - good work on the
>>> quick fixes though!
>>>> 
>>>> Cheers,
>>>> Mike
>>>> 
>>>> On 23/5/02 10:20 PM, "ma...@sm..."
>>> (ma...@sm...) penned the
>>>> words:
>>>> 
>>>>> For WebLogic, using release is the preferred
>>> method. I
>>>>> disagree with your blanket statement. I fixed the
>>>>> release order in the latest CVS and in the branch.
>>> I'm
>>>>> told that it is working correctly on Orion but I'm
>>>>> waiting for official confirmation. I would
>>> appreciate
>>>>> if you could check it with the latest CVS and let me
>>>>> know.
>>>>> 
>>>>> On Thu, 23 May 2002, Mike Cannon-Brookes wrote
>>>>> 
>>>>>> 
>>>>>> Yes - release() is the bane of tag developers.
>>>>>> 
>>>>>> The simple rule is basically - if you're using
>>>>> release, you're doing things
>>>>>> wrong!
>>>>>> 
>>>>>> -mike
>>>>>> 
>>>>>> On 23/5/02 4:28 PM, "Heng Sin Low"
>>>>> (low...@ya...) penned the words:
>>>>>> 
>>>>>>> Ha, finally u got what I mean :)
>>>>>>> 
>>>>>>> Yes, doEndTag should be the right place. I
>>> remember
>>>>>>> Matt have done a massive check in recently that
>>> move
>>>>>>> all this resetting stuff to the release() method.
>>>>>>> 
>>>>>>> Regards,
>>>>>>> Low
>>>>>>> 
>>>>>>>> I'm trying to what out what changes have occurred
>>>>>>> recently, but the taglib
>>>>>>> in CVS is currently broken and not backward
>>>>>>> compatible.
>>>>>>> 
>>>>>>> 
>>>>>>> __________________________________________________
>>>>>>> Do You Yahoo!?
>>>>>>> LAUNCH - Your Yahoo! Music Experience
>>>>>>> http://launch.yahoo.com
>>>>>>> 
>>>>>>> 
>>>>> 
>>> _______________________________________________________________
>>>>>>> 
>>>>>>> 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
>>> 
>>> 
>>> 
>>> _______________________________________________________________
>>> 
>>> 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

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

[Matt B. <ma...@sm...>]

I've checked in a Resin 2.1.1 test case that highlights the NPE problem. If
you click on the test it will work but if you uncomment line 81 in
tests/test-resin.jsp you will get the NPE. I can't see the problem right now
so I committed this test case incase anybody else felt motivated to give it
a look. The test runs fine on other containers so I'm not sure what Resin's
problem is.

-Matt

----- Original Message -----
From: "Matt Baldree" <ma...@sm...>
To: "Sven Kuenzler" <sv...@im...>
Cc: "Webwork-Developer" <web...@li...>
Sent: Monday, May 27, 2002 12:41 PM
Subject: Fw: [Webwork-devel] Re: Webwork 1.0.3 / CVS taglib totally broken?


> It is happening in both places for Resin 2.1.1. so I don't believe it is a
> VS problem. I believe it is a tag problem. I'm looking into it.
>
> ----- Original Message -----
> From: "Sven Kuenzler" <sv...@im...>
> Newsgroups: gmane.comp.java.webwork.devel
> To: <web...@li...>
> Sent: Monday, May 27, 2002 12:13 PM
> Subject: [Webwork-devel] Re: Webwork 1.0.3 / CVS taglib totally broken?
>
>
> > "Mike Cannon-Brookes" <mi...@at...> schrieb im Newsbeitrag
> > news:B9182255.B7E7%mi...@at......
> > > PS More investigating (this time in Resin) since I wrote the above -
> there
> > > are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
> > > ValueStack.java and line 403 of AbstractValueStack.java (both probably
> > > related to null values being looked up and not checked for before
> calling
> > > method on the value object)
> >
> > Strange. Both lines you mention are unchanged to the old ValueStack
> Besides
> > the fact that some of them now live in an abstract class, that is. Are
you
> > sure the NPE's aren't occurring with the old ValueStack as well?
> >
> >     Sven....
> >
> >
> >
> >
> >
> > _______________________________________________________________
> >
> > 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
>
>

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

[Matt B. <ma...@sm...>]

It is happening in both places for Resin 2.1.1. so I don't believe it is a
VS problem. I believe it is a tag problem. I'm looking into it.

----- Original Message -----
From: "Sven Kuenzler" <sv...@im...>
Newsgroups: gmane.comp.java.webwork.devel
To: <web...@li...>
Sent: Monday, May 27, 2002 12:13 PM
Subject: [Webwork-devel] Re: Webwork 1.0.3 / CVS taglib totally broken?


> "Mike Cannon-Brookes" <mi...@at...> schrieb im Newsbeitrag
> news:B9182255.B7E7%mi...@at......
> > PS More investigating (this time in Resin) since I wrote the above -
there
> > are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
> > ValueStack.java and line 403 of AbstractValueStack.java (both probably
> > related to null values being looked up and not checked for before
calling
> > method on the value object)
>
> Strange. Both lines you mention are unchanged to the old ValueStack
Besides
> the fact that some of them now live in an abstract class, that is. Are you
> sure the NPE's aren't occurring with the old ValueStack as well?
>
>     Sven....
>
>
>
>
>
> _______________________________________________________________
>
> 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
>
>

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

[Sven K. <sv...@im...>]

"Mike Cannon-Brookes" <mi...@at...> schrieb im Newsbeitrag
news:B9182255.B7E7%mi...@at......
> PS More investigating (this time in Resin) since I wrote the above - there
> are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
> ValueStack.java and line 403 of AbstractValueStack.java (both probably
> related to null values being looked up and not checked for before calling
> method on the value object)

Strange. Both lines you mention are unchanged to the old ValueStack Besides
the fact that some of them now live in an abstract class, that is. Are you
sure the NPE's aren't occurring with the old ValueStack as well?

    Sven....

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

[Matt B. <ma...@sm...>]

I agree that reset logic should be moved to doEndTag() but this is different
from null'ing attributes. Null'ing attributes should still be in release().
I don't want to make the blanket change at this time. I did fix Property and
Action. I would prefer to make a couple of fixes and all containers should
work. I changed the escaping of outer quotes to true by default. The reason
it was false was just a first guess. I guessed wrong :). I haven't tested
these changes on Resin yet. I can't get to caucho but I will by early this
week.

-Matt

----- Original Message -----
From: "Mike Cannon-Brookes" <mi...@at...>
To: <ma...@sm...>
Cc: <web...@li...>
Sent: Monday, May 27, 2002 2:50 AM
Subject: Re: [Webwork-devel] Webwork 1.0.3 / CVS taglib totally broken?


> Ok - I've checked and from preliminary looks it seems like the
ComponentTag
> is fixed in Orion (haven't tested Resin yet - will do)
>
> PropertyTag:
> I had to fix the PropertyTag again though.
>
> Simple fix: (can't check in myself but here's the details)
> - remove return statement that's not at the end of the doEndTag method)
> - call release() manually before return EVAL_PAGE;
> (see attachment)
>
> Again - the problem here is that data is reset in the release() method
which
> is incorrect ;)
>
> I have a suggested fix for every tag - rename ALL the release() methods to
> reset() and call them manually at the end of the doEndTag() methods (or
> before every return EVAL_PAGE statement) as well. This keeps the 'reset'
> logic compartmentalised but at the same time makes sure that it will be
> called on ALL containers.
>
> I've done this in the attached file so you can see what I mean.
>
> TextUtil:
> Also I changed the default value of "escapeOuterQuotes" to true. Why would
> you ever want this to be false? (Having it false by default means for
> example that any formfield with "text" as a value only gets text which is
v.
> dangerous) Can we change this in CVS? Why was escapeOuterQuotes ever
needed?
>
>
> I'll yell again if I find any more specific problems - but I can see that
> the release() method does things in other tag (which I probably don't use)
> which is bad.
>
> Cheers,
> Mike
>
> PS More investigating (this time in Resin) since I wrote the above - there
> are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
> ValueStack.java and line 403 of AbstractValueStack.java (both probably
> related to null values being looked up and not checked for before calling
> method on the value object)
>
>
> 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 24/5/02 1:03 AM, "ma...@sm..." (ma...@sm...) penned the
> words:
>
> > Alright. Thanks for the reply. We'll get this worked
> > out and commented accordingly. Whatever the outcome,
> > I'll check it against Resin, WL, and Tomcat. I'll leave
> > you and others to check it against Orion. As soon as I
> > get a thumbs up from the community, I'll release a
> > patch 1.0.4.
> >
> > -Matt
> >
> > On Fri, 24 May 2002, Mike Cannon-Brookes wrote
> >
> >>
> >> Matt,
> >>
> >> Thanks mate - I'll check the CVS in the morning and
> > report back. I found a
> >> bunch of changes I needed to make today to get it
> > working, so I'll see how
> >> CVS works for me and start from there again tomorrow
> > (with more mental
> >> energy).
> >>
> >> Re: release() - this is NOT the preferred method at
> > all ;)
> >>
> >> Again (yes, I've checked now with the spec and the
> > author of JSP Tags book)
> >> release() is called by the container to release
> > resources associated with
> >> the tag. It can be called at ANY time the container
> > wishes (you have no
> >> guarantee of when it is called). The only restriction
> > is that it must be
> >> called before the tag instance is garbage collected.
> >>
> >> So - what you have is lazy programmers (WL, Tomcat
> > are the two glaring
> >> examples) who just call release() after every tag
> > invocation - which leads
> >> to people's misplaced belief that it MUST be called
> > everytime. This is not
> >> so! :) Fast containers (like Resin and Orion) pool
> > tag instances and reuse
> >> them (maximising speed), and call release() as few
> > times as possible
> >> (sometimes never). I'm sure it's harder to code it
> > this way, but the end
> >> result is speed.
> >>
> >> Net result:
> >>
> >>    NEVER do anything you want to happen in release()!
> >>
> >> This is true across all the JSP tags in WW and other
> > projects. If you want
> >> logic to occur, or variables to be reset for each tag
> > invocation, reset them
> >> at the end of the doEndTag() method (either before
> > returning, or as a
> >> finally block).
> >>
> >> I'll check CVS in the morning - good work on the
> > quick fixes though!
> >>
> >> Cheers,
> >> Mike
> >>
> >> On 23/5/02 10:20 PM, "ma...@sm..."
> > (ma...@sm...) penned the
> >> words:
> >>
> >>> For WebLogic, using release is the preferred
> > method. I
> >>> disagree with your blanket statement. I fixed the
> >>> release order in the latest CVS and in the branch.
> > I'm
> >>> told that it is working correctly on Orion but I'm
> >>> waiting for official confirmation. I would
> > appreciate
> >>> if you could check it with the latest CVS and let me
> >>> know.
> >>>
> >>> On Thu, 23 May 2002, Mike Cannon-Brookes wrote
> >>>
> >>>>
> >>>> Yes - release() is the bane of tag developers.
> >>>>
> >>>> The simple rule is basically - if you're using
> >>> release, you're doing things
> >>>> wrong!
> >>>>
> >>>> -mike
> >>>>
> >>>> On 23/5/02 4:28 PM, "Heng Sin Low"
> >>> (low...@ya...) penned the words:
> >>>>
> >>>>> Ha, finally u got what I mean :)
> >>>>>
> >>>>> Yes, doEndTag should be the right place. I
> > remember
> >>>>> Matt have done a massive check in recently that
> > move
> >>>>> all this resetting stuff to the release() method.
> >>>>>
> >>>>> Regards,
> >>>>> Low
> >>>>>
> >>>>>> I'm trying to what out what changes have occurred
> >>>>> recently, but the taglib
> >>>>> in CVS is currently broken and not backward
> >>>>> compatible.
> >>>>>
> >>>>>
> >>>>> __________________________________________________
> >>>>> Do You Yahoo!?
> >>>>> LAUNCH - Your Yahoo! Music Experience
> >>>>> http://launch.yahoo.com
> >>>>>
> >>>>>
> >>>
> > _______________________________________________________________
> >>>>>
> >>>>> 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
> >
> >
> >
> > _______________________________________________________________
> >
> > 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
>
>

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

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

Ok - I've checked and from preliminary looks it seems like the ComponentTag
is fixed in Orion (haven't tested Resin yet - will do)

PropertyTag:
I had to fix the PropertyTag again though.

Simple fix: (can't check in myself but here's the details)
- remove return statement that's not at the end of the doEndTag method)
- call release() manually before return EVAL_PAGE;
(see attachment)

Again - the problem here is that data is reset in the release() method which
is incorrect ;) 

I have a suggested fix for every tag - rename ALL the release() methods to
reset() and call them manually at the end of the doEndTag() methods (or
before every return EVAL_PAGE statement) as well. This keeps the 'reset'
logic compartmentalised but at the same time makes sure that it will be
called on ALL containers.

I've done this in the attached file so you can see what I mean.

TextUtil:
Also I changed the default value of "escapeOuterQuotes" to true. Why would
you ever want this to be false? (Having it false by default means for
example that any formfield with "text" as a value only gets text which is v.
dangerous) Can we change this in CVS? Why was escapeOuterQuotes ever needed?


I'll yell again if I find any more specific problems - but I can see that
the release() method does things in other tag (which I probably don't use)
which is bad.

Cheers,
Mike

PS More investigating (this time in Resin) since I wrote the above - there
are now a lot of NPEs due to the new ValueStack stuff. Line 96 of
ValueStack.java and line 403 of AbstractValueStack.java (both probably
related to null values being looked up and not checked for before calling
method on the value object)


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 24/5/02 1:03 AM, "ma...@sm..." (ma...@sm...) penned the
words:

> Alright. Thanks for the reply. We'll get this worked
> out and commented accordingly. Whatever the outcome,
> I'll check it against Resin, WL, and Tomcat. I'll leave
> you and others to check it against Orion. As soon as I
> get a thumbs up from the community, I'll release a
> patch 1.0.4.
> 
> -Matt
> 
> On Fri, 24 May 2002, Mike Cannon-Brookes wrote
> 
>> 
>> Matt,
>> 
>> Thanks mate - I'll check the CVS in the morning and
> report back. I found a
>> bunch of changes I needed to make today to get it
> working, so I'll see how
>> CVS works for me and start from there again tomorrow
> (with more mental
>> energy).
>> 
>> Re: release() - this is NOT the preferred method at
> all ;)
>> 
>> Again (yes, I've checked now with the spec and the
> author of JSP Tags book)
>> release() is called by the container to release
> resources associated with
>> the tag. It can be called at ANY time the container
> wishes (you have no
>> guarantee of when it is called). The only restriction
> is that it must be
>> called before the tag instance is garbage collected.
>> 
>> So - what you have is lazy programmers (WL, Tomcat
> are the two glaring
>> examples) who just call release() after every tag
> invocation - which leads
>> to people's misplaced belief that it MUST be called
> everytime. This is not
>> so! :) Fast containers (like Resin and Orion) pool
> tag instances and reuse
>> them (maximising speed), and call release() as few
> times as possible
>> (sometimes never). I'm sure it's harder to code it
> this way, but the end
>> result is speed.
>> 
>> Net result: 
>> 
>>    NEVER do anything you want to happen in release()!
>> 
>> This is true across all the JSP tags in WW and other
> projects. If you want
>> logic to occur, or variables to be reset for each tag
> invocation, reset them
>> at the end of the doEndTag() method (either before
> returning, or as a
>> finally block).
>> 
>> I'll check CVS in the morning - good work on the
> quick fixes though!
>> 
>> Cheers,
>> Mike
>> 
>> On 23/5/02 10:20 PM, "ma...@sm..."
> (ma...@sm...) penned the
>> words:
>> 
>>> For WebLogic, using release is the preferred
> method. I
>>> disagree with your blanket statement. I fixed the
>>> release order in the latest CVS and in the branch.
> I'm
>>> told that it is working correctly on Orion but I'm
>>> waiting for official confirmation. I would
> appreciate
>>> if you could check it with the latest CVS and let me
>>> know.
>>> 
>>> On Thu, 23 May 2002, Mike Cannon-Brookes wrote
>>> 
>>>> 
>>>> Yes - release() is the bane of tag developers.
>>>> 
>>>> The simple rule is basically - if you're using
>>> release, you're doing things
>>>> wrong!
>>>> 
>>>> -mike
>>>> 
>>>> On 23/5/02 4:28 PM, "Heng Sin Low"
>>> (low...@ya...) penned the words:
>>>> 
>>>>> Ha, finally u got what I mean :)
>>>>> 
>>>>> Yes, doEndTag should be the right place. I
> remember
>>>>> Matt have done a massive check in recently that
> move
>>>>> all this resetting stuff to the release() method.
>>>>> 
>>>>> Regards,
>>>>> Low
>>>>> 
>>>>>> I'm trying to what out what changes have occurred
>>>>> recently, but the taglib
>>>>> in CVS is currently broken and not backward
>>>>> compatible.
>>>>> 
>>>>> 
>>>>> __________________________________________________
>>>>> Do You Yahoo!?
>>>>> LAUNCH - Your Yahoo! Music Experience
>>>>> http://launch.yahoo.com
>>>>> 
>>>>> 
>>> 
> _______________________________________________________________
>>>>> 
>>>>> 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
> 
> 
> 
> _______________________________________________________________
> 
> 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

Fw: [Webwork-devel] ValueStack abstraction

[Matt B. <ma...@sm...>]

Incorporated suggested changes. Look forward to seeing WW work with Cocoon2.

-Matt

----- Original Message -----
From: "Sven Kuenzler" <sv...@im...>
Newsgroups: gmane.comp.java.webwork.devel
To: <web...@li...>
Sent: Tuesday, May 21, 2002 1:08 PM
Subject: [Webwork-devel] ValueStack abstraction


> Some time ago, I told you about integrating Webwork into Cocoon2.
Meanwhile,
> I have mostly finished this task. There is a WebWorkAdapterAction which
> enables Cocoon to act as a WW dispatcher, a WebWorkGenerator that clones
> XSLTServlet's behavior and a XSP taglib that is roughly equivalent to the
> JSP tags. The last component still needs some work (most notably
> implementing some ui:tags) before I can make it publicly available for
> interested parties.
>
> Surprisingly, the integration task was quite easy. I only had to modify
> WebWork itself in one point: WW's ValueStack depends on HttpServletRequest
> and ServletContext. Both of which are generally not available to Cocoon
> components. Instead, they use their own Request et al.  fascades, which
will
> be available even in C2's offline mode.
>
> In order to deal with this, I made a servlet agnostic AbstractValueStack
> which basically does the same thing as the WW original. Except that it
> delegates expression evaluation to  abstract methods when needed. The
> required methods findInContext and getRequestParameter are implemented in
a
> servlet aware ValueStack (and its Cocoon counter part).
>
> I have attached the two source files in question. Modifications are marked
> with SK. My question is whether such an architecture could find its way
into
> the original WW dist. Or is there's even a smarter way to make ValueStack
> useable in a non-servlet context.
>
> The same question goes for ActionContext BTW. In that case, the
modification
> is not strictly necessary, though.
>
>     Sven....
>
>
>
>
>


----------------------------------------------------------------------------
----


>
>
>
>
> _______________________________________________________________
>
> 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
>

[Webwork-devel] Style: CSS for Web Applications

[DeSouza, E. <ede...@ja...>]

Hi,
You guys may want to have a look at Style:
http://style.tigris.org/

Tigris.Org is creating opensource (BSD license) products for developers.

Subversion -- Replacement for CVS (written in C++)
Scarab -- Replacement for Bugzilla (100% Java)
ArgoUML -- UML tool (100% Java)
Axion -- RDBMS (100% Java)
http://www.tigris.org/


bye,
Edwin.

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

[<ma...@sm...>]

Alright. Thanks for the reply. We'll get this worked
out and commented accordingly. Whatever the outcome,
I'll check it against Resin, WL, and Tomcat. I'll leave
you and others to check it against Orion. As soon as I
get a thumbs up from the community, I'll release a
patch 1.0.4.

-Matt

On Fri, 24 May 2002, Mike Cannon-Brookes wrote

> 
> Matt,
> 
> Thanks mate - I'll check the CVS in the morning and
report back. I found a
> bunch of changes I needed to make today to get it
working, so I'll see how
> CVS works for me and start from there again tomorrow
(with more mental
> energy).
> 
> Re: release() - this is NOT the preferred method at
all ;)
> 
> Again (yes, I've checked now with the spec and the
author of JSP Tags book)
> release() is called by the container to release
resources associated with
> the tag. It can be called at ANY time the container
wishes (you have no
> guarantee of when it is called). The only restriction
is that it must be
> called before the tag instance is garbage collected.
> 
> So - what you have is lazy programmers (WL, Tomcat
are the two glaring
> examples) who just call release() after every tag
invocation - which leads
> to people's misplaced belief that it MUST be called
everytime. This is not
> so! :) Fast containers (like Resin and Orion) pool
tag instances and reuse
> them (maximising speed), and call release() as few
times as possible
> (sometimes never). I'm sure it's harder to code it
this way, but the end
> result is speed.
> 
> Net result: 
> 
>    NEVER do anything you want to happen in release()!
> 
> This is true across all the JSP tags in WW and other
projects. If you want
> logic to occur, or variables to be reset for each tag
invocation, reset them
> at the end of the doEndTag() method (either before
returning, or as a
> finally block).
> 
> I'll check CVS in the morning - good work on the
quick fixes though!
> 
> Cheers,
> Mike
> 
> On 23/5/02 10:20 PM, "ma...@sm..."
(ma...@sm...) penned the
> words:
> 
> > For WebLogic, using release is the preferred
method. I
> > disagree with your blanket statement. I fixed the
> > release order in the latest CVS and in the branch.
I'm
> > told that it is working correctly on Orion but I'm
> > waiting for official confirmation. I would
appreciate
> > if you could check it with the latest CVS and let me
> > know.
> > 
> > On Thu, 23 May 2002, Mike Cannon-Brookes wrote
> > 
> >> 
> >> Yes - release() is the bane of tag developers.
> >> 
> >> The simple rule is basically - if you're using
> > release, you're doing things
> >> wrong!
> >> 
> >> -mike
> >> 
> >> On 23/5/02 4:28 PM, "Heng Sin Low"
> > (low...@ya...) penned the words:
> >> 
> >>> Ha, finally u got what I mean :)
> >>> 
> >>> Yes, doEndTag should be the right place. I
remember
> >>> Matt have done a massive check in recently that
move
> >>> all this resetting stuff to the release() method.
> >>> 
> >>> Regards,
> >>> Low
> >>> 
> >>>> I'm trying to what out what changes have occurred
> >>> recently, but the taglib
> >>> in CVS is currently broken and not backward
> >>> compatible.
> >>> 
> >>> 
> >>> __________________________________________________
> >>> Do You Yahoo!?
> >>> LAUNCH - Your Yahoo! Music Experience
> >>> http://launch.yahoo.com
> >>> 
> >>> 
> >
_______________________________________________________________
> >>> 
> >>> 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