date:20040615

Carsten Ziegeler wrote:
Sylvain Wallez wrote: 
 

Makes sense. And migrating some stuff to the env-context may 
help removing some dependencies on Avalon. But whe have to be 
very careful about what attributes we want to move from 
avalon-context to env-context, and be sure they will still 
make sense in a new container and within blocks.

   

Yepp.
 

The only point I'm not sure about is the object model: should 
it be in the env-context, since it also contains that 
env-context? Kind of circular dependency that may be confusing...

   

Yepp, that's true. I would say we leave this in the avalon
context then.
 

If we do that, we can't totally cut the dependency on avalon-context...
What we need actually is some kind of mini-environment that 
would give access only to the "safe" properties of 
o.a.c.e.Environment, i.e. mainly the object model and 
attributes, plus access to the current avalon-context 
attributes through type-safe getter methods.
   

Actually, this mini-environment is the object model. Too bad it's 
currently an untyped Map. Or what about creating an ObjectModel class 
that would be that mini-environment ?

Sylvain
--
Sylvain Wallez  Anyware Technologies
http://www.apache.org/~sylvain   http://www.anyware-tech.com
{ XML, Java, Cocoon, OpenSource }*{ Training, Consulting, Projects }

Re: [VOTE] Unrestricting the FOM

On 15 Jun 2004, at 18:04, Sylvain Wallez wrote:
Jeremy Quinn wrote:

Should I hold back on committing my changes to make ContextHelper 
Contextualizable, to wait to see what comes out of this proposal ?

Wait a bit ;-)
You can still use the ContextAccess class in your own source tree, 
though.
OK

BTW. I think there may be a related issue here with a problem noted a 
while ago, where it was not possible to read parameters setup in 
web.xml as expected using :

org.apache.cocoon.environment.Context.getInitParameter(String 
name)

Unless Cocoon was 'expecting' that parameter to be there.
I don't know if this is a bug or a just a misunderstanding ;)

I guess you made the (sorry!) classical mistake of adding parameters 
to the Cocoon servlet instead of the _webapp context_, which is what 
is accessed using Context.getInitParameter().
Oh no :)

regards jeremy


smime.p7s
Description: S/MIME cryptographic signature

RE: [VOTE] Unrestricting the FOM

Sylvain Wallez wrote: 
> 
> Makes sense. And migrating some stuff to the env-context may 
> help removing some dependencies on Avalon. But whe have to be 
> very careful about what attributes we want to move from 
> avalon-context to env-context, and be sure they will still 
> make sense in a new container and within blocks.
> 
Yepp.

> 
> The only point I'm not sure about is the object model: should 
> it be in the env-context, since it also contains that 
> env-context? Kind of circular dependency that may be confusing...
> 
Yepp, that's true. I would say we leave this in the avalon
context then.

> What we need actually is some kind of mini-environment that 
> would give access only to the "safe" properties of 
> o.a.c.e.Environment, i.e. mainly the object model and 
> attributes, plus access to the current avalon-context 
> attributes through type-safe getter methods.
> 
> >Then in flow, cocoon.context works perfect.
> >  
> >
> 
> Yupp. And with an IOC type2/3 container we could even 
> totally remove the dependency on avalon-context by simply have a
> setContext(o.a.c.e.Context) method ;-)
> 
Yepp :), it would work today if we would move to Fortress.

Carsten

RE: [VOTE] Unrestricting the FOM

Jeremy Quinn wrote: 
> 
> > Currently these attributes are not really used.
> 
> Ah, I though these were the context attributes you could 
> access from FlowScript, as in:
>   cocoon.context.setAttribute ("name", myObject); I have 
> used this before . when I needed to share the same object 
> between several users.
> 
Yes, that's right. I meant Cocoon itself does not use them, it's
up to the users to make sense of them.

> >
> > And we have o.a.a.context.Context which is a map containing
> > many important information about the application.
> 
> Do you mean org.apache.avalon.framework.context.Context ?
> 
Yupp.

> 
> Should I hold back on committing my changes to make ContextHelper 
> Contextualizable, to wait to see what comes out of this proposal ?
> 
Hmm, don't know - we are in development state so we can remove
it if we find a better solution before a release

Carsten

Re: [VOTE] Unrestricting the FOM

Jeremy Quinn wrote:

Should I hold back on committing my changes to make ContextHelper 
Contextualizable, to wait to see what comes out of this proposal ?

Wait a bit ;-)
You can still use the ContextAccess class in your own source tree, though.
BTW. I think there may be a related issue here with a problem noted a 
while ago, where it was not possible to read parameters setup in 
web.xml as expected using :

org.apache.cocoon.environment.Context.getInitParameter(String name)
Unless Cocoon was 'expecting' that parameter to be there.
I don't know if this is a bug or a just a misunderstanding ;)

I guess you made the (sorry!) classical mistake of adding parameters to 
the Cocoon servlet instead of the _webapp context_, which is what is 
accessed using Context.getInitParameter().

Sylvain
--
Sylvain Wallez  Anyware Technologies
http://www.apache.org/~sylvain   http://www.anyware-tech.com
{ XML, Java, Cocoon, OpenSource }*{ Training, Consulting, Projects }

Re: [VOTE] Unrestricting the FOM

Carsten Ziegeler wrote:
Sylvain Wallez wrote:

We already have cocoon.context, couldn't we make things available from there?

Mmmh... don't know if it's a good idea, as the Cocoon core currently doesn't rely on custom attributes in the environment objects (unless I'm mistaken). This would complexify things a bit, IMO.

Hmm, ok we currently have a o.a.c.environment.Context which has some functions from the servlet context object and attributes. Currently these attributes are not really used.

That's what I pointed out above: the Cocoon core currently makes no use
of attributes in env-context, session and request. These are for
application use, and up to now the Cocoon core has not "polluted" these
objects.

And we have o.a.a.context.Context which is a map containing many important information about the application.

Yep. Let's list what's in avalon-context.
The main DefaultContext is initialized with:
- Constants.CONTEXT_ENVIRONMENT_CONTEXT : the environment context (type
o.a.c.e.Context)
- Constants.CONTEXT_WORK_DIR : Cocoon's work directory (type File)
- ContextHelper.CONTEXT_ROOT_URL : the servlet context's URL (type
String - why not URL?)
- Constants.CONTEXT_UPLOAD_DIR : the upload directory (type File)
- Constants.CONTEXT_CACHE_DIR : the cache directory (type File)
- Constants.CONTEXT_CONFIG_URL : location of cocoon.xconf (type URL,
used only to initialize the Cocoon object)
- Constants.CONTEXT_DEFAULT_ENCODING : default form encoding (type String)
- Constants.CONTEXT_CLASS_LOADER : CocoonServlet's classloader (type
ClassLoader, used only to initialize the Cocoon object)
- Constants.CONTEXT_CLASSPATH : the classpath used by the XSP engine
(type String)

The actual Context seen by applications is a ComponentContext that wraps
the above DefaultContext with:
- ContextHelper.CONTEXT_OBJECT_MODEL : the current object model (type Map)
- ContextHelper.CONTEXT_SITEMAP_SERVICE_MANAGER : the current sitemap's
service manager
- ComponentContext.OBJECT_MODEL_KEY_PREFIX : prefix to access members of
the object model (could be removed, as only used in the implementation
of ContextHelper.getRequest()/getResponse())

When in HTTP environment, there's also:
- CocoonServlet.CONTEXT_SERVLET_CONFIG : the ServletConfig
When in portlet environment, there's also
- CocoonPortlet.CONTEXT_PORTLET_CONFIG: the PortletConfig
For me it seems that two contexts is one too much :)

Makes sense. And migrating some stuff to the env-context may help
removing some dependencies on Avalon. But whe have to be very careful
about what attributes we want to move from avalon-context to
env-context, and be sure they will still make sense in a new container
and within blocks.

So, why not combining those somehow. Currently you get the o.a.c.e.Context from the
avalon context.
Now, we could for example store all information we currently have in the avalon context into the environment context and using the environment context as the only reference for application information. If you need this context you can get it from the avalon context. The avalon context only contains this single piece of information (of course for compatibility we let everything remain there as well).

The only point I'm not sure about is the object model: should it be in
the env-context, since it also contains that env-context? Kind of
circular dependency that may be confusing...

What we need actually is some kind of mini-environment that would give
access only to the "safe" properties of o.a.c.e.Environment, i.e. mainly
the object model and attributes, plus access to the current
avalon-context attributes through type-safe getter methods.

Then in flow, cocoon.context works perfect.

Yupp. And with an IOC type2/3 container we could even totally remove
the dependency on avalon-context by simply have a
setContext(o.a.c.e.Context) method ;-)

Sylvain
--
Sylvain Wallez Anyware Technologies
http://www.apache.org/~sylvain http://www.anyware-tech.com
{ XML, Java, Cocoon, OpenSource }*{ Training, Consulting, Projects }

Re: [BUG] forms-lib.js

2004-06-15 Thread Gunnar Brand

Am 16:29 14.06.2004 +0200 schrieb Klaus Bertram:
Hi all,
i test the gui form with the standard IE5.0 based on win2000
there are 2 possible bugs
1 is based in the htmlarea script. (I look at it)
after excluding the load of the script there are a an error in 
forms-lib.js at line 65
the compare of
  if (name == undefined)
can't work because undefined as name is not defined ;)
I add
  var undefined;
at top and it rocks
Tested under IE5.0, IE6.0, Moz 1.6

Klaus
Hello.
Your finding and your solution is correct, IE5.0 (but not Opera, Mozilla 
and IE5.5+) does not know the keyword "undefined".
I found this bug a year ago and used this solution:
eval(navigator.userAgent.indexOf("MSIE 5.0")!=-1 ? "var undefined;" : 
"void(0);");
which doesn't define the variable in browsers that support the keyword (I 
am not talking about IE4 etc here).

IE5.0 is kind of bad for javascript users, because it claims to be 
compatible to a javascript version but then fails to do so.
I can't remember if it is version 1.2 or 1.3 but most likely 1.3.
For example Array methods like push, pop, slice, splice, shift, unshift do 
not exist.
I had quite some fun (not!) to reimplement those methods in pure 
javascript. I don't know about the document.getElementById method, but I 
wrote a simple replacement using document.all for it, too. Of course one 
can't rewrite all missing stuff like the function invoking string replace 
but it was enough to get my application running.

I took a look at this forms-lib.js to check for other errors and there is 
this fixme entry:

  form["forms_submit_id"].value = name;
// FIXME: programmatically submitting the form doesn't trigger 
onsubmit ? (both in IE and Moz)
forms_onsubmit();
form.submit();

Actually, this has been the case since Javascript version 1.1 (and probably 
1.0, but I lack the original netscape documentation of that version) and is 
_correct_ behaviour.
Calling form.submit() should not trigger the onsubmit eventhandler. Buggy 
Opera does so, though.
One can call onsubmit() by hand (after checking if it is defined) or 
click() of one of the submit buttons of the form to trigger the form's 
onsubmit handler. The latter solution is not very safe.
And: Since Opera messes up at this part, relying on onsubmit not being 
invoked when form.submit() is called has also become very dangerous. So 
doing it by hand is probably the best idea.

gunnar

--
G. Brand - interface:projects GmbH
Tolkewitzer Strasse 49
D-01277 Dresden
mail:   [EMAIL PROTECTED]
tel:++49-351-3 18 09 - 41
Ein Unternehmen der interface:business-Gruppe

Re: [VOTE] Unrestricting the FOM

On 15 Jun 2004, at 15:57, Carsten Ziegeler wrote:
Sylvain Wallez wrote:

We already have cocoon.context, couldn't we make things
available from there?

Mmmh... don't know if it's a good idea, as the Cocoon core
currently doesn't rely on custom attributes in the
environment objects (unless I'm mistaken). This would
complexify things a bit, IMO.
Hmm, ok we currently have a o.a.c.environment.Context which
has some functions from the servlet context object and attributes.
This is where I would have expected to find "work-directory", being a 
Servlet-supplied parameter.

Currently these attributes are not really used.
Ah, I though these were the context attributes you could access from 
FlowScript, as in:
	cocoon.context.setAttribute ("name", myObject);
I have used this before . when I needed to share the same object 
between several users.

And we have o.a.a.context.Context which is a map containing
many important information about the application.
Do you mean org.apache.avalon.framework.context.Context ?
For me it seems that two contexts is one too much :)
+ 1
It is rather complicated ;)
So, why not combining those somehow. Currently you get the
o.a.c.e.Context from the avalon context.
Now, we could for example store all information we currently
have in the avalon context into the environment context
and using the environment context as the only reference
for application information. If you need this context
you can get it from the avalon context. The avalon context
only contains this single piece of information (of course
for compatibility we let everything remain there as
well).
Then in flow, cocoon.context works perfect.
WDYT?
YES !!!
Should I hold back on committing my changes to make ContextHelper 
Contextualizable, to wait to see what comes out of this proposal ?

BTW. I think there may be a related issue here with a problem noted a 
while ago, where it was not possible to read parameters setup in 
web.xml as expected using :

org.apache.cocoon.environment.Context.getInitParameter(String name)
Unless Cocoon was 'expecting' that parameter to be there.
I don't know if this is a bug or a just a misunderstanding ;)
regards Jeremy

  If email from this address is not signed
IT IS NOT FROM ME
Always check the label, folks !



smime.p7s
Description: S/MIME cryptographic signature

Bug in HTMLSerializer?

2004-06-15 Thread Stephan Coboos

Hello,
in my XML doc I need to declare some entities. But if I do so, I cant 
use the HTMLSerializer to output my doc because a braket > appears on 
the right top corner. Why? Is it a bug in HTMLSerializer? What can I do 
to avoid this?

My xml doc:

]>

&foo;

My pipeline:

   
   

My browser output (source):
>

bar

Look at the first braket > on top of the source. Why is it there?
Thank you for your help!
Regards
Stephan

RE: [VOTE] Unrestricting the FOM

Sylvain Wallez wrote:
> 
> >We already have cocoon.context, couldn't we make things 
> available from there?
> >  
> >
> 
> Mmmh... don't know if it's a good idea, as the Cocoon core 
> currently doesn't rely on custom attributes in the 
> environment objects (unless I'm mistaken). This would 
> complexify things a bit, IMO.
> 
Hmm, ok we currently have a o.a.c.environment.Context which
has some functions from the servlet context object and attributes.
Currently these attributes are not really used.

And we have o.a.a.context.Context which is a map containing
many important information about the application.

For me it seems that two contexts is one too much :)
So, why not combining those somehow. Currently you get the
o.a.c.e.Context from the avalon context.

Now, we could for example store all information we currently
have in the avalon context into the environment context
and using the environment context as the only reference
for application information. If you need this context
you can get it from the avalon context. The avalon context
only contains this single piece of information (of course
for compatibility we let everything remain there as 
well).
Then in flow, cocoon.context works perfect.

WDYT?

Carsten

RE: excalibur: cocoon dependencies moved!

Antonio Gallardo wrote:
> 
> Yep. Carsten = super programmer!
> 


Thanks, really appreciated, but I think that I'm not better than
most of the others here - far the different. Perhaps I more often
prefer doing over discussing...

> He can manage excalibur stuff, Cocoon and a lot more :-D

> 
> 
> > Looking forward to seeing some cocoon people over on 
> > [EMAIL PROTECTED],
> 
> Thanks, I just few days ago asked Carsten about committer 
> rights for excalibur. I was waiting for the new dev list. I 
> will subscribe now. I hope to be useful there.
> 
> I hope the will be a good atmosphere as in Cocoon. ;-)
> 
:)

Carsten

Re: [VOTE] Unrestricting the FOM

Jeremy Quinn wrote:

I made o.a.c.components.ContextHelper Contextualizable, and added the 
getAvalonContext() method. It all compiles but is not tested from 
FlowScript yet.

I have to get on with some other stuff right now, but hope to have 
this tested and committed by this evening (unless there are objections).

+1!
Sylvain
--
Sylvain Wallez  Anyware Technologies
http://www.apache.org/~sylvain   http://www.anyware-tech.com
{ XML, Java, Cocoon, OpenSource }*{ Training, Consulting, Projects }

Re: Custom Transformer: attributes lost when serialized XML recording

2004-06-15 Thread Vadim Gritsenko

Adam Ratcliffe wrote:
I have a custom transformer that extends AbstractSaxTransformer for
making XML-RPC
calls.  It extracts a method call element from an input document and
sends this to
the remote service for processing.
My problem is that when I use the startSerializedXMLRecording() method
from
AbstractSaxTransformer to collect the method call fragment it records
elements
only and the namespace attribute on the top-level element is lost.
This isn't a problem with the generator that creates the input document
as I can
see the attributes if I serialize the document after generation.
Is there a configuration step that I'm missing?
 

I think you should maintain map of namespace declarations (by processing 
start/end namespace mapping event) and declare all namespaces when you 
start recording. Look around for code samples, I think there is 
something similar in other transformers.

Vadim

public void startElement(String uri, String name, String raw, Attributes
attr)
throws SAXException {
if (name.equals(XMLRPC_REQUEST_ELEMENT)) {
this.stack.push("end");
} else if (name.equals(XMLRPC_SERVICE_ELEMENT)){
   this.startTextRecording(); 
} else if (name.equals(XMLRPC_METHOD_CALL_ELEMENT)) {

this.startSerializedXMLRecording(XMLUtils.defaultSerializeToXMLFormat(tr
ue)); 
} else {
   super.startElement(uri, name, raw, attr);
   }
   }

Re: Using HTTPS as a src for a file generator

2004-06-15 Thread Steven Noels

On 14 Jun 2004, at 22:56, Steven Noels wrote:
It sure was - I'll Wikify my success tomorrow for posterity.
http://wiki.cocoondev.org/Wiki.jsp?page=HTTPsSources
Thanks!

--
Steven Noelshttp://outerthought.org/
Outerthought - Open Source Java & XMLAn Orixo Member
Read my weblog athttp://blogs.cocoondev.org/stevenn/
stevenn at outerthought.orgstevenn at apache.org

Re: [VOTE] Unrestricting the FOM

On 15 Jun 2004, at 11:27, Jeremy Quinn wrote:
On 15 Jun 2004, at 10:57, Sylvain Wallez wrote:
Jeremy Quinn wrote:
On 15 Jun 2004, at 10:11, Sylvain Wallez wrote:

Looking at the vote results, the general opinion is to remove the 
API restrictions (got only +1's),

Good
but not tie the FOM to a particular Avalon object (got lots of +0's 
and a -1).

This is more difficult to understand, but I could see why some were 
worried.

So let's drop this cocoon.avalonContext proposal. Firstly because 
it becomes less useful if the FOM is unrestricted, and secondly 
because we can easily add the ContextAccess I proposed to Jeremy as 
a utility class. People using that class will know by doing so that 
their app becomes tied to Avalon.

That is fine by me.
Shall I commit the sample class you provided?
Do you have any preferences for where it should live?
Or do you have a more sophisticated way of adding this 
functionality, that you would prefer to use instead?

I see two possible locations for this code:
- in a new o.a.c.components.flow.util.AvalonContextAccessor class 
(note the explicit mention to Avalon to clearly show the dependency 
on the framework)
I can understand this one
- in o.a.c.components.ContextHelper, which alread holds a number of 
context-related methods.
I am not quite sure how this functionality would get added to 
o.a.c.components.ContextHelper.

It currently has all static methods, which are passed an existing 
org.apache.avalon.framework.context.Context Object.

The second solution has the benefit of concentrating all 
Context-related features in a single class and so has my preference.
Are you suggesting that this class be made Contextualizable and be 
given a getContext() method ?

So to use it for getting a Context, you would call
cocoon.createObject (ContextHelper).getContext ()
on it?
OK
I made o.a.c.components.ContextHelper Contextualizable, and added the 
getAvalonContext() method. It all compiles but is not tested from 
FlowScript yet.

I have to get on with some other stuff right now, but hope to have this 
tested and committed by this evening (unless there are objections).

regards Jeremy


  If email from this address is not signed
IT IS NOT FROM ME
Always check the label, folks !



smime.p7s
Description: S/MIME cryptographic signature

Re: [VOTE] Unrestricting the FOM

On 15 Jun 2004, at 10:57, Sylvain Wallez wrote:
Jeremy Quinn wrote:
On 15 Jun 2004, at 10:11, Sylvain Wallez wrote:

Looking at the vote results, the general opinion is to remove the 
API restrictions (got only +1's),

Good
but not tie the FOM to a particular Avalon object (got lots of +0's 
and a -1).

This is more difficult to understand, but I could see why some were 
worried.

So let's drop this cocoon.avalonContext proposal. Firstly because it 
becomes less useful if the FOM is unrestricted, and secondly because 
we can easily add the ContextAccess I proposed to Jeremy as a 
utility class. People using that class will know by doing so that 
their app becomes tied to Avalon.

That is fine by me.
Shall I commit the sample class you provided?
Do you have any preferences for where it should live?
Or do you have a more sophisticated way of adding this functionality, 
that you would prefer to use instead?

I see two possible locations for this code:
- in a new o.a.c.components.flow.util.AvalonContextAccessor class 
(note the explicit mention to Avalon to clearly show the dependency on 
the framework)
I can understand this one
- in o.a.c.components.ContextHelper, which alread holds a number of 
context-related methods.
I am not quite sure how this functionality would get added to 
o.a.c.components.ContextHelper.

It currently has all static methods, which are passed an existing 
org.apache.avalon.framework.context.Context Object.

The second solution has the benefit of concentrating all 
Context-related features in a single class and so has my preference.
Are you suggesting that this class be made Contextualizable and be 
given a getContext() method ?

So to use it for getting a Context, you would call
cocoon.createObject (ContextHelper).getContext ()
on it?
Thanks
regards Jeremy


smime.p7s
Description: S/MIME cryptographic signature

Re: [VOTE] Unrestricting the FOM

Jeremy Quinn wrote:
On 15 Jun 2004, at 10:11, Sylvain Wallez wrote:

Looking at the vote results, the general opinion is to remove the API 
restrictions (got only +1's),

Good
but not tie the FOM to a particular Avalon object (got lots of +0's 
and a -1).

This is more difficult to understand, but I could see why some were 
worried.

So let's drop this cocoon.avalonContext proposal. Firstly because it 
becomes less useful if the FOM is unrestricted, and secondly because 
we can easily add the ContextAccess I proposed to Jeremy as a utility 
class. People using that class will know by doing so that their app 
becomes tied to Avalon.

That is fine by me.
Shall I commit the sample class you provided?
Do you have any preferences for where it should live?
Or do you have a more sophisticated way of adding this functionality, 
that you would prefer to use instead?

I see two possible locations for this code:
- in a new o.a.c.components.flow.util.AvalonContextAccessor class (note 
the explicit mention to Avalon to clearly show the dependency on the 
framework)
- in o.a.c.components.ContextHelper, which alread holds a number of 
context-related methods.

The second solution has the benefit of concentrating all Context-related 
features in a single class and so has my preference.

WDYT?
Sylvain
--
Sylvain Wallez  Anyware Technologies
http://www.apache.org/~sylvain   http://www.anyware-tech.com
{ XML, Java, Cocoon, OpenSource }*{ Training, Consulting, Projects }

Re: [VOTE] Unrestricting the FOM

On 15 Jun 2004, at 10:11, Sylvain Wallez wrote:
Reinhard Poetz wrote:
Sylvain Wallez wrote:
Hi all,
More and more, the limitations of objects provided by the FOM seem 
like arbitrary constraints that go in the way of people and produce 
confusion. Furthermore, these restrictions only apply to the JS 
flowscript and not to JavaFlow, thus making JS flowscript a second 
zone citizen compared to Java code.

That's why I propose to remove these restrictions my having 
cocoon.request, cocoon.response and cocoon.context providing the 
full API defined by the interfaces in org.apache.cocoon.environment.

Furthermore, I propose to add cocoon.avalonContext to provide access 
to the various data held by this object.

More background on this subject is available in the "Less is more... 
or less" discussion [1] and my answer to Jeremy's problem [2] that 
shows how simple it is to workaround the restricted FOM.

Please cast your votes!
- [+1] to remove restrictions on existing objects.
- [?] to add cocoon.avalonContext.

I'm not sure about avalonContext.`What happens if we really drop 
Avalon as base framework? Do statements containing 
cocoon.avalonContext still work or will they be hidden (because 
Flowscript is interpreted) broken? If the user has to use the 
workaround and she recompiles the code with Cocoon 3.0 and the 
compiler doesn't compile her code anymore, she knows that she has a 
problem. Or will a possible legacy mode hide that she may have a 
problem?

We should really try to make moving from Cocoon 2.x to 3.x as smooth 
as possible and users that only use Flowscripts and the sitemap 
shouldn't be forced to change their applications if they want to use 
Cocoon in the same way as they have been used to do.

Looking at the vote results, the general opinion is to remove the API 
restrictions (got only +1's),
Good
but not tie the FOM to a particular Avalon object (got lots of +0's 
and a -1).
This is more difficult to understand, but I could see why some were 
worried.

So let's drop this cocoon.avalonContext proposal. Firstly because it 
becomes less useful if the FOM is unrestricted, and secondly because 
we can easily add the ContextAccess I proposed to Jeremy as a utility 
class. People using that class will know by doing so that their app 
becomes tied to Avalon.

That is fine by me.
Shall I commit the sample class you provided?
Do you have any preferences for where it should live?
Or do you have a more sophisticated way of adding this functionality, 
that you would prefer to use instead?

Thanks for your help, and thanks everyone for your votes.
regards Jeremy



smime.p7s
Description: S/MIME cryptographic signature

Re: [VOTE] Unrestricting the FOM

Reinhard Poetz wrote:
Sylvain Wallez wrote:
Hi all,
More and more, the limitations of objects provided by the FOM seem 
like arbitrary constraints that go in the way of people and produce 
confusion. Furthermore, these restrictions only apply to the JS 
flowscript and not to JavaFlow, thus making JS flowscript a second 
zone citizen compared to Java code.

That's why I propose to remove these restrictions my having 
cocoon.request, cocoon.response and cocoon.context providing the full 
API defined by the interfaces in org.apache.cocoon.environment.

Furthermore, I propose to add cocoon.avalonContext to provide access 
to the various data held by this object.

More background on this subject is available in the "Less is more... 
or less" discussion [1] and my answer to Jeremy's problem [2] that 
shows how simple it is to workaround the restricted FOM.

Please cast your votes!
- [+1] to remove restrictions on existing objects.
- [?] to add cocoon.avalonContext.

I'm not sure about avalonContext.`What happens if we really drop 
Avalon as base framework? Do statements containing 
cocoon.avalonContext still work or will they be hidden (because 
Flowscript is interpreted) broken? If the user has to use the 
workaround and she recompiles the code with Cocoon 3.0 and the 
compiler doesn't compile her code anymore, she knows that she has a 
problem. Or will a possible legacy mode hide that she may have a problem?

We should really try to make moving from Cocoon 2.x to 3.x as smooth 
as possible and users that only use Flowscripts and the sitemap 
shouldn't be forced to change their applications if they want to use 
Cocoon in the same way as they have been used to do.

Looking at the vote results, the general opinion is to remove the API 
restrictions (got only +1's), but not tie the FOM to a particular Avalon 
object (got lots of +0's and a -1).

So let's drop this cocoon.avalonContext proposal. Firstly because it 
becomes less useful if the FOM is unrestricted, and secondly because we 
can easily add the ContextAccess I proposed to Jeremy as a utility 
class. People using that class will know by doing so that their app 
becomes tied to Avalon.

Sylvain
--
Sylvain Wallez  Anyware Technologies
http://www.apache.org/~sylvain   http://www.anyware-tech.com
{ XML, Java, Cocoon, OpenSource }*{ Training, Consulting, Projects }

Re: Custom widgets in CForms

2004-06-15 Thread Vilya Harvey

Scott Yeadon wrote:
Hello,
Has anyone tried to create custom widgets within the Forms framework? 
Beyond the business logic of their validation rules, etc, I cannot find 
the code which actually creates the widgets visible user interface? 
Anyone got a how-to on any of this?
This recently created page on the cocoon wiki may help: 
http://wiki.cocoondev.org/Wiki.jsp?page=CocoonFormsCreatingWidgets

Vil.
--
__
   o|   _. /  \|o._  _  _ ._  _  ._  _ _|_
\/ ||\/(_|| (|/||| |(/_(_)| |(/_o| |(/_ |_
 / \__
http://website.lineone.net/~vilya

Re: Looking for a HowTo on adding widgets to CForms.

Andreas Hochsteger wrote:
Hi Vilya,
I'm sorry that I can't point you to an existing HOWTO.
I had the same problem some time ago when I wanted to write a new data type
for CForms.
Perhaps it's time to collect this information on the wiki.
I've created two pages which can be filled by those who know something about
these things or are forced to learn it (hint!):
http://wiki.cocoondev.org/Wiki.jsp?page=CocoonFormsCreatingWidgets
 

I modified this page a bit, mainly about modifying the 
WidgetReplacingPipe which is clearly something that's not needed for 
most of the widgets.

Sylvain
--
Sylvain Wallez  Anyware Technologies
http://www.apache.org/~sylvain   http://www.anyware-tech.com
{ XML, Java, Cocoon, OpenSource }*{ Training, Consulting, Projects }

Re: [VOTE] Unrestricting the FOM

2004-06-15 Thread Reinhard Poetz

Sylvain Wallez wrote:
Hi all,
More and more, the limitations of objects provided by the FOM seem 
like arbitrary constraints that go in the way of people and produce 
confusion. Furthermore, these restrictions only apply to the JS 
flowscript and not to JavaFlow, thus making JS flowscript a second 
zone citizen compared to Java code.

That's why I propose to remove these restrictions my having 
cocoon.request, cocoon.response and cocoon.context providing the full 
API defined by the interfaces in org.apache.cocoon.environment.

Furthermore, I propose to add cocoon.avalonContext to provide access 
to the various data held by this object.

More background on this subject is available in the "Less is more... 
or less" discussion [1] and my answer to Jeremy's problem [2] that 
shows how simple it is to workaround the restricted FOM.

Please cast your votes!
- [+1] to remove restrictions on existing objects.
- [?] to add cocoon.avalonContext.

I'm not sure about avalonContext.`What happens if we really drop Avalon 
as base framework? Do statements containing cocoon.avalonContext still 
work or will they be hidden (because Flowscript is interpreted) broken? 
If the user has to use the workaround and she recompiles the code with 
Cocoon 3.0 and the compiler doesn't compile her code anymore, she knows 
that she has a problem. Or will a possible legacy mode hide that she may 
have a problem?

We should really try to make moving from Cocoon 2.x to 3.x as smooth as 
possible and users that only use Flowscripts and the sitemap shouldn't 
be forced to change their applications if they want to use Cocoon in the 
same way as they have been used to do.

WDYT?
--
Reinhard

Re: Custom widgets in CForms