Once I made a component like this.
here is demonstration
http://ruzin.fei.tuke.sk:8080/majcher/app
the source code should be here, http://knopp.sk/modal.zip
but a) it's wicket 1.1
b) there could be some glitches on java side
Once I finish the tree I might rework it to make the modal
I think you should add Button(s) to the form and add onclick attribute
modifier on them (or right on to the markup input type=submit
wicket:id=submit onlick=return confirm('...');/ )
-Matej
Dorel Vaida wrote:
I'm using wicket 1.2 final and trying to use the form's 'onsubmit'
handler to
returning false in the form's onsubmit is the
standard way of elegantly preventing the form from being submitted.
Eelco
On 7/17/06, Matej Knopp [EMAIL PROTECTED] wrote:
I think you should add Button(s) to the form and add onclick attribute
modifier on them (or right on to the markup input type
What URL do you exactly display in the iframe? I think the behavior is
caused by PageMap in 1.2 which behaves as stack. So to get over this you
have to put the page that is shown in the iframe to separate pagemap.
You can change the pagemap in which wicket is created by either
specifying it in
=66D2B1310B47081B62EE203A648CDAE8?wicket:interface=:0:statusForm:data:nodeGroups:0:deprovision::IBehaviorListenerwicket:behaviorId=0',
function() { }, function() { });
2. Can you please explain what is th purpose of PageMap in this case?
Thanks
-Nili
Matej Knopp wrote:
What URL do you exactly
Actually, I've been thinking about this hybrid urls for a while (also
discussed it with igor) and I think it's something we should have. I'm
probably going to refactor current request encoding/decoding stuff so I
plan to address this there.
-Matej
Igor Vaynberg wrote:
the
David Leangen wrote:
Hello!
so what you want is instead of url doing this:
/mytree
click
/app?wicket:interface=
to do this
/mytree
click
/mytree?wicket:interface=
is this correct? and if so...
Yes, that's exactly correct!
i proposed this awhile back, and i think we
In wicket 2.0 you can create pages that are not stored in session
(stateless pages).
Anyway, It's not possible with wicket to store state of tree in url, and
I doubt it ever will be. URL length is limited so it wouldn't work for
bigger trees. And wicket is managing state for whole pages, so it
I see. With wicket 2.0 it should be possible to build a page like this.
With 1.2 it's not. I guess you can make your own tree subclassing
WicketTree and replace links on nodes with bookmarkable links, but that
won't really help you since the page will still be stored in session.
-Matej
David
Look at DataView from wicket-extensions.
-Matej
[EMAIL PROTECTED] Imam wrote:
Hello users,
I wanted to know if wicket has support for using a component similar to
PagingNavigation to tell it to selectively
bring the items it will view in a Page depending on the page link clicked.
My
response.
is there any way around it? what happens if i don't lock?
Matej Knopp wrote:
Ittay Dror wrote:
this is from RequestCycle:
private final void processEventsAndRespond()
{
// Use any synchronization lock provided
Anyway, I don't really understand, why is the ejb object actually being
serialized. Do you store your service objects in session?
-Matej
Igor Vaynberg wrote:
well, the problem might be that it is serialized by wicket itself.
this is done because you have the logger set to debug to help
a reference to
it in
http session so I can recall that exact instance back from the
server
later. This is how it was done in the app that is currently
running
in production on JBoss.
On 7/6/06, Matej Knopp [EMAIL PROTECTED] wrote:
Anyway, I don't really understand, why is the ejb object
You get the error because you are using CompoundPropertyModel and has
the security manager in tomcat turned on.
CompoundPropertyModel uses introspection to set / get properties and for
doing that, it needs to run with permission
java.lang.reflect.ReflectPermission suppressAccessChecks
You can
Ittay Dror wrote:
this is from RequestCycle:
private final void processEventsAndRespond()
{
// Use any synchronization lock provided by the target
Object lock = getRequestTarget().getLock(this);
if (lock != null)
{
Doesn't setDefaultFormProcessing(false) work for you? The values should
be preserved even if model is not update. At least if you're using
wicket 1.2 or newer.
-Matej
Alexei Sokolov wrote:
I need some help with the following scenario.
I have a form and when a user is halfway through it,
No, label should behave just like any other component. Perhaps you
should check if the wicket:id of label matches your object's property.
And make sure you don't specify the compound model to the label
directly, but you specify it to it's parent.
-Matej
samyem wrote:
Label does not seem to
Why do you have to specify object as component model to be able to use
it in onSubmit?
-Matej
cowwoc wrote:
Hi,
I'd like to pass in a non-String model into a Button so I can use it
for onSubmit() but then the Button text ends up being crap (hashcode of
the model object). I can
Hi, I actually haven't done that yet, although it is on my to-do list.
Hopefully in couple of weeks I'll look into it.
-Matej
Nili Adoram wrote:
Dear Mr. Knopp,
On February you have posted a wonderful example of an ajax modal dialog.
Did you upgrade this example to wicket-1.2?
Thanks for
Strange. I'm looking into PageMapChecker#renderHead and it doesn't seem
obvious why the generated url doesn't contain pageMap name.
Are you using a custom URL encoding scheme or anything like that?
-Matej
Ittay Dror wrote:
Hi,
I'm using firefox. when i duplicate a tab, the tab reloads in
Hmm.. The problem with this kind of bugs is that it's very difficult to
reproduce.
IMHO The problem is in this line:
else {document.location.href =
'/page/SomePage;jsessionid=E43E09C7B61190F3C8C5E98CB988F21F?';}
The url should contain pagemap name (something like wicket-1).
In wicket-1.2 the
to the
encoding strategy to encode the path. in case the target was not
mounted, it would have encoded the url itself (in my case, the target
is IBookmarkablePageRequestTarget)
the code there is less than trivial, should i copy paste? can i
reference it somehow?
ittay
Matej Knopp wrote:
Hmm
This is going to cause you problem with multiple windows. You should
rather obtain the pagemap using Page#getPageMap.
Anyway, this doesn't feel like the right way to get to previous page.
Can't you just propagate the page instance and then use
setResponsePage(previousPageInstance) where
Yeah, with hibernate 3 the HibernateTemplate doesn't really makes much
sense. With hibernate 2 it was good for translating exceptions, since
HibernateException was checked.
However, we've been working on a JDBC project lately and Spring's
JdbcTemplate was a huge win to have.
-Matej
Igor
IOC + AOP + Remoting + Lot of other stuff.
Spring is a swiss army knife of web development :)
-Matej
Igor Vaynberg wrote:
i dont think so. spring is an ioc container at the very least. obix is
just a lib to make it easy to read in config files.
-Igor
On 6/14/06, *Vincent Jenks *
That are not two web application. Just two different wicket applications
(wicket servlet) in the same webapplication (context).
There is no way of sharing session between two web applications.
-Matej
Vincent Jenks wrote:
I was wondering this myself actuallyhow can a session be shared
...it seems like a good candidate for this portal project for
externalizing resources. In other words - Spring might be overkill if
it can be done more easily w/ Obix if I were to use EJB3.
I guess I was thinking aloudheh.
On 6/14/06, Matej Knopp [EMAIL PROTECTED] wrote:
IOC + AOP
Well, this seems to be a side effect of current wicket multi-tab/window
support.
The easiest workaround I can think of is to disable the multi-tab
support using
getPageSettings().setAutomaticMultiWindowSupport(false);
Could you try if that helps?
-Matej
Gianugo Rabellino wrote:
Hi there,
There is jasper report integration project in wicket-stuff. I just don't
know if it's up-to-date with 1.2.
-Matej
[EMAIL PROTECTED] wrote:
Hi All
we are devloping web application using wicket ,jasper report, hibernate
I want to know how to call jasper reports in wicket
Thankas
Check your imports (on the beginning of your java code). It seems like
you are importing pc.LoginPage instead of mobile.LoginPage.
-Matej
R.A wrote:
Hi.
I create two packages and some class having same class name.
mobile.LoginPage
mobile.MainPage
pc.LoginPage
pc.MainPage
In
setGender(int) { br
* ... br
* } br
* br
* int getGender() { br
* ... br
* } br
*
* /code
*
* @author Matej Knopp
*
*/
public class SimpleSelection implements Serializable {
ListInteger ids;
ListSimpleSelectionItem items
Although this is intended to be used for simple constants drop downs.
But I guess similar approach could be used if you store id's of your
domain objects in models.
-Matej
Matej Knopp wrote:
It's doable, but you have to write your own choice renderer.
I've made a SimpleSelection class
),
* render.getChoicesModel(),
* renderer));
*
* ...
*
* public int getPersonId() {
* ...
* }
*
* public void setPersonId() {
* ...
* }
* /pre
*
* @author Matej Knopp
*
* @param ItemType type of item (entity)
* @param IdType type
just call model.getObject()
Just a thought...
On Wed, 2006-05-31 at 22:35 +0200, Matej Knopp wrote:
:) Yeah, getObject is a little confusing at the beginning. The parameter
is mostly used in compound models (CompoundPropertyModel) where the
model is shared between multiple
No, you're not doing anything wrong. SVN on sourceforget seems to be down.
-Matej
Chris M wrote:
Trying to checkout wicket 1.2 fails. Using the command
svn checkout
https://svn.sourceforge.net/svnroot/wicket/branches/WICKET_1_2/wicket
wicket
results in the following:
svn: PROPFIND
that, a lot
of problems I had where suddenly easy to solve :)
--
greetings from Berlin,
RĂ¼diger Schulz
Matej Knopp wrote on 01.06.2006 at 18:46:
Plus those methods are redundant. It would just clutter the
interface.
The contract is well defined. Maybe javadoc of get
Well, that's a limitation of current wicket form handling. It doesn't
support partial updates out of box. If you need behavior like this you
have to implement the form processing on your own.
-Matej
Gianugo Rabellino wrote:
Might be a wicked use case, but I can't find an elegant way of making
Hi.
The problem is that you would need more constructors, and some wicket
components (DropDownChoice)... have already lot of constructors.
And the concept is currently limited to RepeatingView. As far as I know
there are no other components where you can use this, are they?
-Matej
Timo
Isn't your model object null? Try to set the model object to a different
value, it should help.
-Matej
Dorel Vaida wrote:
When I reach with the selection in a combo box, to the first item in the
list, wicket places a Choose one option as the first item in the
combo. Am I doing something
I you want to change the displayed string just put
youFormId.yourDropDownChoiceId.null=My null String
in your page's property file.
-Matej
Matej Knopp wrote:
Isn't your model object null? Try to set the model object to a different
value, it should help.
-Matej
Dorel Vaida wrote:
When I
new RestartResponseException(...) should do it.
-Matej
Vincent Jenks wrote:
How can I simply stop a page from rendering and redirect to another?
Say I've got a condition where I've found an object to be null or an
unacceptable value and I'd rather send the user to another page w/o
rendering
I don't know why but I just don't like the idea of long-running
transactions. I think web applications are just too unpredictable to
hold one transaction during multiple request. You can never be sure when
(and if) the next request will come.
OpenSessionInView filter has always be sufficient
I'm not sure I understand you but you load the collections separately?
Can't you just do something like this?
class MyPage extends Page {
public MyPage() {
IModel m1 = new LoadableDetachableModel() {
Object load() {
return [load collection 1];
}
}
add(new
) and doing calculations against it - for each
iteration in the ListView loop.
In other words, I just need access to the raw object data in it's real
data-type after I've loaded it into a detachable model - my question
is simply that, how do I get it back to its original type.
On 5/31/06, Matej
perfect. The part I was unsure about was what to
pass for the arguement for getObject()
Thanks Matej!
On 5/31/06, Matej Knopp [EMAIL PROTECTED] wrote:
Suppose you have
IModel model = new LoadableDetachableModel() { ...
you can get the model any time:
ListItem myItems = (ListItem
someone
can correct me where I'm making false assumptions.
On 5/31/06, Matej Knopp [EMAIL PROTECTED] wrote:
I don't know why but I just don't like the idea of long-running
transactions. I think web applications are just too unpredictable to
hold one transaction during multiple request. You can never
No, it can't. There's no support for multipart forms ajax submit currently.
-Matej
Ayodeji Aladejebi wrote:
Is it possible to submit a Multi-Form for File Upload with
AjaxSubmitButton?
Please Just Yes or No and I will be fine
can Ajax handle multipart-form submit because i want to update a
As far as I know, it is not possible to upload file using ajax. If you
need to achieve similar effect, consider using hidden iframe for this.
-Matej
Ayodeji Aladejebi wrote:
the desgn initiative of wicket is awesome and quite tempting for
developers you know :) ...Many see it as another SDK
of wicket currently
-Igor
On 5/29/06, Ayodeji Aladejebi [EMAIL PROTECTED] wrote:
in Wicket or in the whole WWW ?
On 5/29/06, Matej Knopp [EMAIL PROTECTED] wrote:
No, it can't. There's no support for multipart forms ajax submit
currently.
-Matej
Ayodeji Aladejebi wrote
I guess this is because in 1.2 pagemap behaves like stack. Johan?
-Matej
Ari Suutari wrote:
Hi,
I'm having a problem with wicket's ajax stuff where an ajax request
fired by previous page hits the web server after user has clicked a link
which has already transferred him to next page.
This
.
- Original Message -
From: Matej Knopp [EMAIL PROTECTED] mailto:[EMAIL PROTECTED]
To: wicket-user@lists.sourceforge.net
mailto:wicket-user@lists.sourceforge.net
Sent: Monday, May 22, 2006 11:44 AM
Subject: Re: [Wicket-user] pagemap problem with late arriving ajax
request
The problem is that wicket-ajax doesn't support multipart forms.
-Matej
Nick Heudecker wrote:
I'm attempting to upload a file using a form with an AjaxSubmitButton.
Ideally, I'd like to swap out a panel once the form is submitted, but
I'm not getting that far. When I submit the form, I'm
Eelco Hillenius wrote:
The only thing to take care of is not to allow IE6 to work in
standard-compliance mode. So far, I had no problems with this setup. I'm
doing quite complicated layouts using just css (no tables) and
everything works fine. I just hope that IE7 won't screw everything up.
solve so many problems we have...
So you let IE stay in quircks mode and have for firefox that
-moz-box-sizing: border-box;
and for other browsers (maybe IE7) the CSS3? standard: box-sizing:
border-box;
What is your doctype?
johan
On 5/18/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL
Bruno Borges wrote:
One thing at a time:
1) setVisible works, when not doing through Ajax components like AjaxLink;
2) ListView outputs one different id for each repeated line.
3) when calling ListView.setVisible(false) from some
onClick(AjaxRequestTarget), it does not works.
So pay
On 5/18/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Bruno Borges wrote:
One thing at a time:
1) setVisible works, when not doing through Ajax components like
AjaxLink;
2) ListView outputs one different id for each repeated line
Michiel Trimpe wrote:
Has anyone already tried to implement LDAP-authentication with Wicket?
Wicket has nothing to do with authentication. You might consider using
Acegi or JAAS for this.
Altough Acegi seems little over-complicated.
Hmm.. thinking of it, it would be nice to have a wicket
Igor Vaynberg wrote:
allow myself to quote...myself
|| imho, html is the best layout manager out there for browser apps. add
css to the mix and you have a great skin manager as well.
i never said css was great for layout manager :) and yes the box model
is broken.
The box model really is
Cool. It took an entire day to deliver this mail :)
Anyway, I've seen that this is already solved in current head
(markupparser got new method), so just ignore this.
-Matej
Matej Knopp wrote:
done.
http://sourceforge.net/tracker/index.php?func=detailaid=1488809group_id=119783atid=684978
done.
http://sourceforge.net/tracker/index.php?func=detailaid=1488809group_id=119783atid=684978
-Matej
Eelco Hillenius wrote:
Wow, this was a big thread. But now it seems to have died quitely.
Could someone please fetch conclusions and put this in an RFE?
Eelco
Martijn Dashorst wrote:
Add the updating region to the ajax target of the button request. That
should update the URL for the self updating timer.
This doesen't really make sense, versioning is turned of during ajax
requests.
Martijn
On 5/11/06, *Johan Compagner * [EMAIL PROTECTED]
Ittay Dror wrote:
public final Object getModelObject()
{
final IModel model = getModel();
if (model != null)
{
// If this component has the root model for a compound model
if (getFlag(FLAG_HAS_ROOT_MODEL))
{
// we
But why would you do that? this flag is set automatically when you set a
compound model to a component.
-Matej
Ittay Dror wrote:
i want to overcome the issue which i've just posted about, and to set
the FLAG_HAS_ROOT_MODEL to false, but the value is private. can it be
made protected?
I solved this some time ago with own preprocessor that replaces
${key} with the value from property files.
I know this is not very wicket-like, but the markup looks much simpler
with ${key} then with wicket:message key=key/wicket:message
-Matej
Igor Vaynberg wrote:
why not? in a preview you
, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
I solved this some time ago with own preprocessor that replaces
${key} with the value from property files.
I know this is not very wicket-like, but the markup looks much simpler
with ${key} then with wicket:message key=key
Johan Compagner wrote:
still the question remains how random is random ;)
right. maybe one shared atomically increased integer would do better job?
On 5/9/06, *Igor Vaynberg* [EMAIL PROTECTED]
mailto:[EMAIL PROTECTED] wrote:
just checked it into core
the param looks like this
, but... if Wicket is going to have ${key}... ${something}
in htmls, let's go back to JSP 2.0 and use EL... just a thought
On 5/9/06, *Matej Knopp * [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
So far I can say it works well for me.
The only problematic (or cumbersome) thing was that I had
/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Bruno Borges wrote:
This is completely different. ${key} is simply replaced by a string
from
property file. No EL, no expressions, no ognl, nothing. Just a simple
string.
Comparing wicket:message key=key
Juergen Donnerstag wrote:
On 5/9/06, Matej Knopp [EMAIL PROTECTED] wrote:
$${key}? Or whatever you want.
I don't need this behavior to be in wicket core. I'm much more
interested in clean and simple preprocessing filters, something wicket
lacks currently.
Unfortunately IMarkupFilter does
, *Martijn Dashorst* [EMAIL PROTECTED]
mailto:[EMAIL PROTECTED] wrote:
On 5/9/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED]
wrote:
Johan Compagner wrote:
still the question remains how random is random ;)
right. maybe one shared atomically increased
Matej Knopp wrote:
Juergen Donnerstag wrote:
On 5/9/06, Matej Knopp [EMAIL PROTECTED] wrote:
$${key}? Or whatever you want.
I don't need this behavior to be in wicket core. I'm much more
interested in clean and simple preprocessing filters, something wicket
lacks currently.
Oh, stupid me
LocalizedMarkupParser#initFilterChain
method (I derived my own parser from MarkupParser).
At least protected Markup MarkupParser.getMarkup() would help.
-Matej
Matej Knopp wrote:
Matej Knopp wrote:
Juergen Donnerstag wrote:
On 5/9/06, Matej Knopp [EMAIL PROTECTED] wrote:
$${key}? Or whatever you
the hierarchy yet? maybe this can wait until 2.0.
-Igor
On 5/9/06, * Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Well, it looks that this is not as easy as it seemed to be.
I've made a MarkupFilter that translates the string. But I need
to give it ContainerInfo
a lot of api breaks,
but if you compare code in 1.2 to 1.1 they can almost be different
frameworks.
-Igor
On 5/9/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
If WicketMessageTagHandler can get resource.getContainerInfo(), why my
filter can't?
There's
PROTECTED] wrote:
the flipside is of course that we were able to refactor a ton without
breaking too much api. you can argue that 1.2 has a lot of api breaks,
but
if you compare code in 1.2 to 1.1 they can almost be different
frameworks.
-Igor
On 5/9/06, Matej Knopp [EMAIL PROTECTED] wrote
Eelco Hillenius wrote:
On 5/9/06, Matej Knopp [EMAIL PROTECTED] wrote:
Igor Vaynberg wrote:
I don't mind API breaks. Not at all. If I touch code that is not a part
of Stable API, I take the risk of having to change my code when wicket
version changes. I really don't mind.
Well, not everyone
:)
-Matej
Matej Knopp wrote:
Eelco Hillenius wrote:
On 5/9/06, Matej Knopp [EMAIL PROTECTED] wrote:
Igor Vaynberg wrote:
I don't mind API breaks. Not at all. If I touch code that is not a part
of Stable API, I take the risk of having to change my code when wicket
version changes. I really don't
Heh, that's funny, of course you didn't, I did.
-Matej
Eelco Hillenius wrote:
On 5/9/06, Matej Knopp [EMAIL PROTECTED] mailto:[EMAIL PROTECTED]
wrote:
Igor Vaynberg wrote:
I don't mind API breaks. Not at all. If I touch code that is not
a part
of Stable
done.
Johan Compagner wrote:
can you report a bug for this?
On 4/29/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Johan Compagner wrote:
we are trying to insert values (the model object) in the string
you ask for.
you could use method getString
:
On 4/29/06, Matej Knopp [EMAIL PROTECTED] wrote:
Johan Compagner wrote:
this is pretyt much all in place.
I don't believe in a cookie and or url state what is that?
storing a
page in an url?
We have a branch where we have a first draft of ClientSide Page
saving
bookmarkable page links for things like
tabs. But it also destroys part of Wicket's programming model.
Eelco
On 4/30/06, Matej Knopp [EMAIL PROTECTED] wrote:
Eelco Hillenius wrote:
Yeah there's just no perfect world as long as browsers work the way
they work. The big, really big advantage of client
On 4/30/06, Eelco Hillenius [EMAIL PROTECTED] wrote:
At the moment you e.g. click on links of the pageable list component,
you'll be increasing versions - unless that pageable list is
unversioned - and your tab links will thus point to older versions.
Eelco
On 4/30/06, Matej Knopp [EMAIL
Eelco Hillenius wrote:
I'm not convinced this is a good solution, but I'm concerned that
most people are ignoring this problem. I love everything about
wicket, except this. In this regard, wicket is a giant step
backwards from other frameworks like webwork, struts, etc.
Most struts-like apps
Well, every framework that manages the session state I know has problems
with keeping client in sync with server. Tracking browser windows/tabs
is difficult and unreliable. Wicket compared to the alternatives
currently does a very good (though hardly perfect) job.
The partial solution could
Hi,
Either Component.getString() is broken or I'm doing something wrong.
Say I have a class Item with one property name.
I have a form (wicket:id=myForm) with CompoundPropertyModel(new Item());
Then I call form.getString(someStringKey)
I end with an exception, because getString calls
is the Root object of that model? then
model.getObject(null) should be called
sounds right.
-Matej
johan
On 4/29/06, * Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Hi,
Either Component.getString() is broken or I'm doing something wrong.
Say I have a class
Johan Compagner wrote:
this is pretyt much all in place.
I don't believe in a cookie and or url state what is that? storing a
page in an url?
We have a branch where we have a first draft of ClientSide Page saving
(in an javascript variable that is then set in a hidden field of all the
Guess this won't be enough, you have to redraw also the node that was
selected previously.
What I did was that I redrawn the whole tree.
It was a 10 minutes hack, I took wicket tree and changed all links to
AjaxFailbackLink.
I can say that even redrawing the whole tree was magnitudes better
pool.
Maybe this isn't any more such a big problem for most folks.
It occurs me when updating a label with 100 ms timer (well, actually,
maybe the timer in AjaxSelfUpdatingTimerBehavour shouldn't fire
until previous request has completed ?)
Ari S.
- Original Message - From: Matej Knopp
folks.
It occurs me when updating a label with 100 ms timer (well, actually,
maybe the timer in AjaxSelfUpdatingTimerBehavour shouldn't fire
until previous request has completed ?)
Ari S.
- Original Message - From: Matej Knopp [EMAIL PROTECTED]
To: wicket-user@lists.sourceforge.net
in AjaxSelfUpdatingTimerBehavour shouldn't fire
until previous request has completed ?)
Ari S.
- Original Message - From: Matej Knopp [EMAIL PROTECTED]
To: wicket-user@lists.sourceforge.net
Sent: Wednesday, April 12, 2006 9:51 PM
Subject: Re: [Wicket-user] wicket ajax memory leak with IE
Hi, I made
Johan Compagner wrote:
can you send me the wicket-ajax.js file that you use?
So that i have exactly the right version?
johan
On 4/12/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Hi, I made some changes, this really helps, at least it works for me:
// AJAX
Hi.
I changed the wicket-ajax.js to create only one instance of
XmlHttpRequest, and the memory leak in IE seemed to vanish.
So the problem is that creating XmlHttpRequest probably causes a memory
leak. Howewer, having only one instance of XmlHttpRequest doesn't seem
to be the best idea
Ari Suutari wrote:
Hi,
Thanks for you response,
I changed the wicket-ajax.js to create only one instance of
XmlHttpRequest, and the memory leak in IE seemed to vanish.
So the problem is that creating XmlHttpRequest probably causes a
memory leak. Howewer, having only one instance of
Ari Suutari wrote:
need several requests at the same time. So maybe pooling of
XmlHttpRequest(s) could work...
Btw, does dojo use pooling (since there was no leak with it) ?
Maybe it would make sense to check how they are getting around
this.
Ari S.
Don't know, might be worth
Yeah, it is. However, it's much smaller. But still, 10-15 kb per request.
-Matej
Igor Vaynberg wrote:
is the leak still there if you turn off the ajax debug mode in
IAjaxSettings?
-Igor
On 4/12/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Ari Suutari wrote
)
-Matej
Igor Vaynberg wrote:
i looked through the javascript but nothing jumped out at me. did you
notice anything cause im stumped.
-Igor
On 4/12/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Yeah, it is. However, it's much smaller. But still, 10-15 kb per
because ms cant get it right.
-Igor
On 4/12/06, *Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
AFAIK There's nothing wrong with javascript. The problem is that
creating XMLHttpRequest every time causes memory leak in IE.
People tend to reuse one instance
-urlencoded);
transport.send(body);
return true;
}
Matej Knopp wrote:
I guess I have seen one. But it shouldn't be that difficult to get it
done. I guess I might be able to look into it tonight.
-Matej
Igor Vaynberg wrote:
is there a pooling implementation somewhere i can take a look
, not
innerHtml)
Vaynberg wrote:
what exactly is the problem? ajax requests use the same settings, so if
you have stripwickettags=false both full and ajax renders should act the
same.
-Igor
On 3/19/06, * Matej Knopp* [EMAIL PROTECTED] mailto:[EMAIL PROTECTED] wrote:
Hi, there seems
401 - 500 of 686 matches
Mail list logo