Smalltalk › Frameworks & Tools › Seaside › Seaside General

liveCallback and SUAutocompleter

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

20 messages Options

Burella Juan M.

liveCallback and SUAutocompleter

Hi, I have a problem with Scriptaculous and SeasideAsync. I have a liveCalback ( on a select tag ) when gets evaluated it shows a SUAutocompleterTest instance. Then when I select an item, all components are gone except the SUAutocompleterTest. On the other hand, if I switch/replace the SUAutocompleterTest and render another component it works fine.
I´ve attached an example to reproduce the behavior.
Any ideas?

    aRenderer select
                    list: #(one two);
                    selected: #one;
                    callback:[:f | ];
                    liveCallback: [:anObject :h | h div id: 'xxlive';with: [
                                anObject = #two ifTrue: [ h render: component "a SUAutocompleterTest instance"]]
                            ].
    aRenderer div id: 'xxlive';with: ''

- I add SULibrary and WAAsyncScripts libraries
- I'm using Squeak3.8/Seaside2.7a1-mb.155/SeasideAsync-mb.48.

Regards.
Juan Matias.

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Lukas Renggli

Re: liveCallback and SUAutocompleter

> Hi, I have a problem with Scriptaculous and SeasideAsync. I have a
> liveCalback ( on a select tag ) when gets evaluated it shows a
> SUAutocompleterTest instance. Then when I select an item, all components are
> gone except the SUAutocompleterTest. On the other hand, if I switch/replace
> the SUAutocompleterTest and render another component it works fine.
> I´ve attached an example to reproduce the behavior.
> Any ideas?
>
> aRenderer select
> list: #(one two);
> selected: #one;
> callback:[:f | ];
> liveCallback: [:anObject :h | h div id: 'xxlive';with: [
> anObject = #two ifTrue: [ h render:
> component "a SUAutocompleterTest instance"]]
> ].
> aRenderer div id: 'xxlive';with: ''

I doubt that SeasideAsync and Scriptaculous work together that close.
I would try to use Scriptaculous **xor** SeasideAsync.

Lukas

--
Lukas Renggli
http://www.lukas-renggli.ch
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Michel Bany

Re: liveCallback and SUAutocompleter

In reply to this post by Burella Juan M.

>
> - I add SULibrary and WAAsyncScripts libraries
> - I'm using Squeak3.8/Seaside2.7a1-mb.155/SeasideAsync-mb.48.
>
Hi Juan,
Before giving up, you may want to give it a try with the latest
SeasideAsync (53).
It has some interesting enhancements that may help in your case.
Michel.
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Michel Bany

Re: liveCallback and SUAutocompleter

In reply to this post by Burella Juan M.

>
> - I add SULibrary and WAAsyncScripts libraries
> - I'm using Squeak3.8/Seaside2.7a1-mb.155/SeasideAsync-mb.48.
>
Hi Juan,
You may want to give it a try with the latest SeasideAsync (53).
It has some interesting enhancements that may help in your case.
Michel.
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Burella Juan M.

Re: liveCallback and SUAutocompleter

Hi Michel
I've tested it in Seaside2.7a1-dc.128/SeasideAsync-mb.54 but doesn't work fine, the problem persist. What enhacements should I look for in order to fix it?
thanks for your answer

On 11/18/06, Michel Bany <[hidden email]> wrote:

>
> - I add SULibrary and WAAsyncScripts libraries
> - I'm using Squeak3.8/Seaside2.7a1-mb.155/SeasideAsync-mb.48.
>
Hi Juan,
You may want to give it a try with the latest SeasideAsync (53).
It has some interesting enhancements that may help in your case.
Michel.
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Michel Bany

Re: liveCallback and SUAutocompleter

On 20 Nov 2006, at 16:49 , Burella Juan M. wrote:

> Hi Michel
> I've tested it in Seaside2.7a1-dc.128/SeasideAsync-mb.54 but
> doesn't work fine, the problem persist. What enhacements should I
> look for in order to fix it?
> thanks for your answer
>

Looks like a problem in Seaside and/or Scriptaculous.

The SeasideAsync javascripts use the XML parser supplied by the
XMLHttpRequest object. Therefore the response from the live
callback must be well-formed XML, otherwise, it cannot be parsed
(at least in some browsers).

When the response from the live callback includes some Scriptaculous
stuff, then it is not well-formed, because it contains some ampersands.
Since Seaside generates the script tags with #html: rather than #text:
there is no encoding of ampersands into entities. This is not well-
formed
XML and cannot be parsed, at least by the Firefox XML parser.

I do not know what should be fixed, Seaside or Scriptaculous.

Lukas, any idea ?

Meanwhile, you can try the following fix to SUAjax :

parameters
^ (self options at: 'parameters' ifAbsentPut:
[ SUJoin join: OrderedCollection new with: '&' ]) collection

HTH,
Michel.

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Lukas Renggli

Re: Re: liveCallback and SUAutocompleter

> I do not know what should be fixed, Seaside or Scriptaculous.
>
> Lukas, any idea ?
>
> Meanwhile, you can try the following fix to SUAjax :
>
> parameters
> ^ (self options at: 'parameters' ifAbsentPut:
> [ SUJoin join: OrderedCollection new with: '&' ]) collection

Don't try this, it breaks Scriptaculous. The encoding works fine as
long as you only attach your scripts to element events, such as
#onClick:, #onChange:, etc.

The problem is that the method #with: in WAScriptTag was implemented
in a very strange way, it bypassing the encoding and caused the
observed problem. The missing thing was the <![CDATA[ ... ]]> tag.
Unfortunately the same bug is duplicated in at least 4 places trough
out the framework. The latest version of 2.7 fixes these problems. I
don't know the effect of this change, it might well be that older
browser have their problems. I tested it on Safari and FireFox.

Cheers,
Lukas

--
Lukas Renggli
http://www.lukas-renggli.ch
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Philippe Marschall

Re: Re: liveCallback and SUAutocompleter

2006/11/21, Lukas Renggli <[hidden email]>:

> > I do not know what should be fixed, Seaside or Scriptaculous.
> >
> > Lukas, any idea ?
> >
> > Meanwhile, you can try the following fix to SUAjax :
> >
> > parameters
> > ^ (self options at: 'parameters' ifAbsentPut:
> > [ SUJoin join: OrderedCollection new with: '&' ]) collection
>
> Don't try this, it breaks Scriptaculous. The encoding works fine as
> long as you only attach your scripts to element events, such as
> #onClick:, #onChange:, etc.
>
> The problem is that the method #with: in WAScriptTag was implemented
> in a very strange way, it bypassing the encoding and caused the
> observed problem. The missing thing was the <![CDATA[ ... ]]> tag.
> Unfortunately the same bug is duplicated in at least 4 places trough
> out the framework. The latest version of 2.7 fixes these problems. I
> don't know the effect of this change, it might well be that older
> browser have their problems. I tested it on Safari and FireFox.

Don't use CDATA ffs. It does simply not work which is why we don't do
it for text. We encode text properly so let's also encode scripts
properly.

Philippe
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Lukas Renggli

Re: Re: Re: liveCallback and SUAutocompleter

> Don't use CDATA ffs. It does simply not work which is why we don't do
> it for text. We encode text properly so let's also encode scripts
> properly.

Unfortunately encoding does not work, this was what I first tried of
course. CDATA works and validates in the browser I tested.

Lukas

--
Lukas Renggli
http://www.lukas-renggli.ch
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Philippe Marschall

Re: liveCallback and SUAutocompleter

In reply to this post by Lukas Renggli

2006/11/21, Lukas Renggli <[hidden email]>:

> > I do not know what should be fixed, Seaside or Scriptaculous.
> >
> > Lukas, any idea ?
> >
> > Meanwhile, you can try the following fix to SUAjax :
> >
> > parameters
> > ^ (self options at: 'parameters' ifAbsentPut:
> > [ SUJoin join: OrderedCollection new with: '&'
> ]) collection
>
> Don't try this, it breaks Scriptaculous. The encoding works fine as
> long as you only attach your scripts to element events, such as
> #onClick:, #onChange:, etc.
>
> The problem is that the method #with: in WAScriptTag was implemented
> in a very strange way, it bypassing the encoding and caused the
> observed problem. The missing thing was the <![CDATA[ ... ]]> tag.
> Unfortunately the same bug is duplicated in at least 4 places trough
> out the framework. The latest version of 2.7 fixes these problems. I
> don't know the effect of this change, it might well be that older
> browser have their problems. I tested it on Safari and FireFox.

Wouldn't the same behavior also apply to styles?

Philippe
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Lukas Renggli

Re: Re: liveCallback and SUAutocompleter

> Wouldn't the same behavior also apply to styles?

I don't know, but that is rarely used in Seaside anyway.

The only thing I found out was that a script that has the & encoded as
& doesn't work within a <script>, but does within an attribute
value. JavaScript code only works in a <script> when the contents is
not encoded, but this turns the XHTML invalid if the contents is not
within a CDATA construct. The CDATA construct should not harm any
browser that does not understand it, so what is your aversion against
it except that you don't like it?

Lukas

--
Lukas Renggli
http://www.lukas-renggli.ch
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Boris Popov, DeepCove Labs (SNN)

Re: Re: Re: liveCallback and SUAutocompleter

In reply to this post by Burella Juan M.

Re: Re: Re: [Seaside] liveCallback and SUAutocompleter

BTW, scriptaculous docs actually suggest wrapping all scripts in CDATA to avoid issues with using < and > inside, so its can't be *that* much of a problem. I recall seeing the suggestion other times too.

Cheers!

-Boris
(Sent from a BlackBerry)

----- Original Message -----
From: [hidden email] <[hidden email]>
To: The Squeak Enterprise Aubergines Server - general discussion. <[hidden email]>
Sent: Tue Nov 21 13:35:42 2006
Subject: Re: Re: Re: [Seaside] liveCallback and SUAutocompleter

> Don't use CDATA ffs. It does simply not work which is why we don't do
> it for text. We encode text properly so let's also encode scripts
> properly.

Unfortunately encoding does not work, this was what I first tried of
course. CDATA works and validates in the browser I tested.

Lukas

--
Lukas Renggli
http://www.lukas-renggli.ch
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Philippe Marschall

Re: Re: liveCallback and SUAutocompleter

In reply to this post by Lukas Renggli

2006/11/21, Lukas Renggli <[hidden email]>:
> > Wouldn't the same behavior also apply to styles?
>
> I don't know, but that is rarely used in Seaside anyway.
>
> The only thing I found out was that a script that has the & encoded as
> & doesn't work within a <script>,

This is really strange.

> but does within an attribute
> value. JavaScript code only works in a <script> when the contents is
> not encoded, but this turns the XHTML invalid if the contents is not
> within a CDATA construct. The CDATA construct should not harm any
> browser that does not understand it, so what is your aversion against
> it except that you don't like it?

CDATA is not part of HTML. It's part of XML and we serve as text/html
which means it gets treated as HTML. It's just plain wrong.
Second, it breaks as soon as somewhere in a script is ']]>' lets say
for XML parsing or whatever. For the person this happens this will
mean real pain. It's not a solution, it just appears to works in some
cases.

Philippe
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Michel Bany

Re: liveCallback and SUAutocompleter

In reply to this post by Burella Juan M.

Michel Bany

Re: Re: liveCallback and SUAutocompleter

In reply to this post by Philippe Marschall

but does within an attribute
value. JavaScript code only works in a <script> when the contents is
not encoded, but this turns the XHTML invalid if the contents is not
within a CDATA construct. The CDATA construct should not harm any
browser that does not understand it, so what is your aversion against
it except that you don't like it?

CDATA is not part of HTML. It's part of XML and we serve as text/html
which means it gets treated as HTML. It's just plain wrong.
Second, it breaks as soon as somewhere in a script is ']]>' lets say
for XML parsing or whatever. For the person this happens this will
mean real pain. It's not a solution, it just appears to works in some
cases.

See http://javascript.about.com/library/blxhtml.htm

This article seems to say that CDATA should be used in XHTML for

javascript. XHTML is what is targeted by Seaside anyway.

Also, I believe that the w3c html 4 spec mentions CDATA.

Finally, I found out that the mistake was also in SeasideAsync itself

which was not properly encoding the scripts produced by Scriptaculous.

So I changed SeasideAsync to enclose the scripts within a CDATA

without the comment constructs. The comment construct is not used

because SeasideAsync callbacks respond with a 100% well-formed

XML document (including the <?xml . . . ?> header) and the CDATA

cannot be parsed when it is polluted by the comment construct.

My initial tests indicate that Scriptaculous scripts can now be generated

from within SeasideAsync callbacks. Who is going to try a SeasideAsync

live request from within a Scriptaculous callback ? How deep can we

nest the SeasideAsync and Scriptaculous callbacks ?

Cheers,

Michel.

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Lukas Renggli

Re: Re: Re: liveCallback and SUAutocompleter

In reply to this post by Philippe Marschall

> > The only thing I found out was that a script that has the & encoded as
> > & doesn't work within a <script>,
>
> This is really strange.

Yes, agreed. Feel free to try it out yourself.

> > but does within an attribute
> > value. JavaScript code only works in a <script> when the contents is
> > not encoded, but this turns the XHTML invalid if the contents is not
> > within a CDATA construct. The CDATA construct should not harm any
> > browser that does not understand it, so what is your aversion against
> > it except that you don't like it?
>
> CDATA is not part of HTML. It's part of XML and we serve as text/html
> which means it gets treated as HTML. It's just plain wrong.

So what? Luckily there is a Javascript comment around the CDATA so
non-xml parsers still handle it properly. Maybe then the mime-type of
responses should be changed to text/xml, this would be more consistent
anyway?

> Second, it breaks as soon as somewhere in a script is ']]>' lets say
> for XML parsing or whatever. For the person this happens this will
> mean real pain. It's not a solution, it just appears to works in some
> cases.

Please find a better solution that works instead of only complaining.

Lukas

--
Lukas Renggli
http://www.lukas-renggli.ch
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Philippe Marschall

Re: Re: Re: liveCallback and SUAutocompleter

2006/11/22, Lukas Renggli <[hidden email]>:
> > > The only thing I found out was that a script that has the & encoded as
> > > & doesn't work within a <script>,
> >
> > This is really strange.
>
> Yes, agreed. Feel free to try it out yourself.

If it wasn't you I wouldn't believe it until I tried it.
Digging deeper there is CDATA and PCDATA in html :(
Not to be confused with <![CDATA[

> > > but does within an attribute
> > > value. JavaScript code only works in a <script> when the contents is
> > > not encoded, but this turns the XHTML invalid if the contents is not
> > > within a CDATA construct. The CDATA construct should not harm any
> > > browser that does not understand it, so what is your aversion against
> > > it except that you don't like it?
> >
> > CDATA is not part of HTML. It's part of XML and we serve as text/html
> > which means it gets treated as HTML. It's just plain wrong.
>
> So what? Luckily there is a Javascript comment around the CDATA so
> non-xml parsers still handle it properly.

Ah, I missed this point completly. This changes a lot. Acutally this
addresses everything but a script that contains ']]>'

> Maybe then the mime-type of
> responses should be changed to text/xml, this would be more consistent
> anyway?
>
> > Second, it breaks as soon as somewhere in a script is ']]>' lets say
> > for XML parsing or whatever. For the person this happens this will
> > mean real pain. It's not a solution, it just appears to works in some
> > cases.
>
> Please find a better solution that works instead of only complaining.

Yeah I know. And if I have a problem with the VM I should just stfu and fix it.

All I can think of right now would break on html parsers.

Philippe
_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Lukas Renggli

Re: Re: Re: Re: liveCallback and SUAutocompleter

> > > > but does within an attribute
> > > > value. JavaScript code only works in a <script> when the contents is
> > > > not encoded, but this turns the XHTML invalid if the contents is not
> > > > within a CDATA construct. The CDATA construct should not harm any
> > > > browser that does not understand it, so what is your aversion against
> > > > it except that you don't like it?
> > >
> > > CDATA is not part of HTML. It's part of XML and we serve as text/html
> > > which means it gets treated as HTML. It's just plain wrong.
> >
> > So what? Luckily there is a Javascript comment around the CDATA so
> > non-xml parsers still handle it properly.
>
> Ah, I missed this point completly. This changes a lot. Acutally this
> addresses everything but a script that contains ']]>'

Yes, that special thing of encoding will be tackled as soon as someone
is complaining. Different encoding within different parts of the
resulting page is something that is neglected right now, you can only
choose between no-encoding or HTML encoding. Looks like a project to
properly implement the different encodings for URLs, Scripts, CSS,
Attributes, etc.

The attached script is a good start to play around with these encoding
issues (in the context of scripts for now). It should dispay 3 alert
boxes with the string "done ...", should not raise warnings/errors and
should validate.

> All I can think of right now would break on html parsers.

Yeah, text/xml breaks almost all applications as Web browsers get very
picky and display all kinds of warnings and errors.

Cheers,
Lukas

--
Lukas Renggli
http://www.lukas-renggli.ch

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

WAScriptEncodingTest.st (1K) Download Attachment

Burella Juan M.

Re: liveCallback and SUAutocompleter

In reply to this post by Michel Bany

It worked fine.
Thanks very much for all answers.

Juan Matías.

On 11/22/06, Michel Bany <[hidden email]> wrote:

On 20 Nov 2006, at 16:49 , Burella Juan M. wrote:

> Hi Michel
> I've tested it in Seaside2.7a1-dc.128/SeasideAsync-mb.54 but
> doesn't work fine, the problem persist. What enhacements should I
> look for in order to fix it?
> thanks for your answer

Now that Lukas has enhanced Seaside and Scriptaculous to provide
proper encoding for the script elements, I also enhanced SeasideAsync
in the same way. You may want to try again with the latest combination
- Seaside2.7a1-lr.131
- Scriptaculous-lr.154
- SeasideAsync-mb.56
I tried it with your code and got the autocompleter to display within
the
rest of the page, fully functionnal.

Best luck,
Michel.

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside

Michel Bany

Re: liveCallback and SUAutocompleter

On 22 Nov 2006, at 16:08 , Burella Juan M. wrote:

> It worked fine.
> Thanks very much for all answers.
>
> Juan Matías.

Thanks for the feedback.

Be aware that there are some new issues in using SeasideAsync
because of the new CDATA encoding for javascripts.
This is being looked at.

Cheers,
Michel.

_______________________________________________
Seaside mailing list
[hidden email]
http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside