gmane.comp.java.dwr.user

Sylvie Villard added you as a friend on Facebook

Facebook — 2008-05-17T11:58:07

Sylvie added you as a friend on Facebook. We need to confirm that you know Sylvie in order for you to be friends on Facebook. To confirm this friend request, follow the link below: http://www.facebook.com/n/?reqs.php Thanks, The Facebook Team ___ Want to control which emails you receive from Facebook? Go to: http://www.facebook.com/editaccount.php?notifications&md=ZnJpZW5kO2Zyb209MTA2NjgyMDQ4ODt0bz03MTgwMzI2OTM=

Re: Serializing closures

Joe Walker — 2008-05-17T10:59:40

I wonder how easy it would be to rip the GWT compiler out of GWT and use it at runtime? I guess it works on source, rather than bytecode. Decompiler? Joe. On Sat, May 17, 2008 at 11:48 AM, Jose Noheda public.gmane.org> wrote:

Re: basic authentication (the right way)

Joe Walker — 2008-05-17T10:57:43

The reason we've never exposed this is that DWR doesn't guarantee to use XHR. If you have an old Safari (and maybe a new Safari too), IE6 on Windows server, are using file-upload, cross-domain, or reverse ajax, then DWR may use iframe or script tags, where the username and password don't work the same way if at all. You can still: - set a custom header / cookie / parameter and check in an AjaxFilter - use cookie based auth like EE role based security - add the auth details to an extra parameter to the called methods. Joe. 2008/5/17 Johan Hoogenboezem public.gmane.org>:

Re: Serializing closures

Jose Noheda — 2008-05-17T10:48:01

I can also imagine a lot of uses for an outbound converter with a "define" method so new code can be injected to the page. Ideally something like GWT that converts Java to Javascript code automatically? On Fri, May 16, 2008 at 6:15 PM, Joe Walker public.gmane.org> wrote:

basic authentication (the right way)

Johan Hoogenboezem — 2008-05-17T10:14:45

Hi, We’re using basic authentication for our GI app. Before we started using DWR, we passed login credentials via the open() method on GI’s jsx3.net.Request object. With DWR, I expected something similar to be required. However, we ended up hacking engine.js like this: if (DWREngine.USER_NAME != undefined) { batch.req.open(batch.httpMethod, request.url, batch.async, DWREngine.USER_NAME, DWREngine.USER_PASSWORD); } else batch.req.open(batch.httpMethod, request.url, batch.async); We would set DWREngine.USER_NAME and DWREngine.USER_PASSWORD after logging in successfully. This works, but I’ve never felt comfortable with it. It’s messy etc. etc. etc. What is the recommended way for doing this? Something like options to be set for the engine? Regards Johan Hoogenboezem No virus found in this outgoing message. Checked by AVG. Version: 7.5.524 / Virus Database: 269.23.16/1445 - Release Date: 2008/05/15 07:25 PM

Re: memory and performance fixes for 2.0.4

Joe Walker — 2008-05-17T09:19:03

Duh! - you're right. Maybe the best solution is to have session clearing on a best-attempts basis, leaving out old IEs. Joe.

RE: Using Interface as method parameter

Mike Wilson — 2008-05-16T20:37:49

Great, all aboard! Ok, I'll compile a list of the suggestions I have collected and complement it with some very rough design suggestions, and we can take the discussion from there. I'll post it as a new thread next week. A nice weekend to all of you! Mike _____ From: joseph.walker-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org [mailto:joseph.walker-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org] On Behalf Of Joe Walker Sent: den 16 maj 2008 18:00 To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: Re: [dwr-user] Using Interface as method parameter Also backlogged, but also interested. Joe. On Fri, May 16, 2008 at 2:43 PM, Randy Jones public.gmane.org> wrote: I'm definitely in. I haven't got to this yet, but still very high on my list. Randy From: Mike Wilson [mailto:mikewse-PkbjNfxxIARBDgjK7y7TUQ< at >public.gmane.org] Sent: Thursday, May 15, 2008 6:12 PM To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter [going through my backlog, sorry for the delay] Hi Randy, I have collected some different requests on the class-mapping stuff, and your dilemma overlaps partly with another request from some time ago. Maybe we should have a discussion about these with Joe and maybe Jose? If either of you are interested in contributing to this stuff then we could open up a design discussion in a new mail thread? Best regards Mike _____ From: Randy Jones [mailto:randy_jones-v48TKWNzwFw< at >public.gmane.org] Sent: den 9 april 2008 15:08 To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter Mike, The JS library I'm using is build on top of dojo so it is using all the inheritance infrastructure dojo provides. My problem is I have a situation where I want to pass an object to the server, via DWR of course, to then do some processing. An example that very well explains my situation, or so I hope. My JS library has an interface called Shape and Rectangle and Circle are concrete classes that inherit from Shape. I would like to have my Java method accept Shape instead of Rectangle or Circle. Originally when I started this project I tried to just pass the object, but I had bad luck b/c DWR marshaled a lot of extra information (handlers etc). So I then went to essentially cloning them using the JS constructors as you mentioned. This was all fine when I was only really working with Rectangles, but not they want Circles too (insert normal developer frustration based on floating requirements). So the JS library has toJSON() functions on all the objects so it's easy for me to just the value representation of an object (no handlers or internal vars) which would then be a great object for me to pass to DWR with the exception of this interface issue. I haven't looked at this part of DWR really, if you think you have a way around this or think this might be a good idea to look at implementing, if you point me in the right direction I'll see what I can work up. Randy From: Mike Wilson [mailto:mikewse-PkbjNfxxIARBDgjK7y7TUQ< at >public.gmane.org] Sent: Wednesday, April 09, 2008 3:47 AM To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter Hi Randy, No, as the class mapping system uses the JavaScript class, objects need to be created with "new " to participate in class mapping. Note though that DWR already does this with outbound objects so all objects you receive from the server will already have the correct class association. How are your client objects created as you hint that you have no control over this? If you have no control over object creation I guess there is only one concrete class to map against, and then maybe you could use the concrete class (instead of interface) as argument in your Java API? Best regards Mike _____ From: Randy Jones [mailto:randy_jones-v48TKWNzwFw< at >public.gmane.org] Sent: den 8 april 2008 22:11 To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter Mike, I have that working for a single case, as your example describes, but I'm looking to expand, so is there also a way of defining what class it is w/o using the JS Constructor? I'd prefer this b/c basically what I'm doing right now is similar to Joe's work with GI where I'm creating Java classes that represent a JS library so you can see I'd rather not have to basically clone the existing objects. To continue with you're example something like: myObj.className = "testdwrclasses.Concrete" I thought I remembered a discussion on something like this but I couldn't find it in the archives, thus prompting my inquiry for documentation I couldn't discover. Randy From: Mike Wilson [mailto:mikewse-PkbjNfxxIARBDgjK7y7TUQ< at >public.gmane.org] Sent: Tuesday, April 08, 2008 5:31 AM To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter Randy, I'm sorry, I know that this should have been done a long time ago and it is on my to-do list. Though, like most people I need to do paid work at day-time, and all my free time has been devoted to DWR memory and performance fixes lately. Anyway, this should get you going: 1.Make sure you have converters for both the interface and applicable concrete classes. 2.Add a "javascript" attribute to these converters. Example (dwr.xml): You may then also create new concrete objects from the client through generated JavaScript constructors: var c = new Concrete; Best regards Mike _____ From: Randy Jones [mailto:randy_jones-v48TKWNzwFw< at >public.gmane.org] Sent: den 4 april 2008 23:48 To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: [dwr-user] Using Interface as method parameter Mike/Joe, I actually have a need to use an interface in a method parameter and couldn't find documentation on it, is it somewhere on the site? Randy

RE: memory and performance fixes for 2.0.4

Mike Wilson — 2008-05-16T20:32:41

I think the timeout used for the 30ms canceling would be blocked by us hanging in the sync call, right? (But IE8 actually has a new XHR timeout setting so maybe the days of safe sync calls are coming in a near future :-) Best regards Mike _____ From: joseph.walker-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org [mailto:joseph.walker-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org] On Behalf Of Joe Walker Sent: den 16 maj 2008 18:17 To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: Re: [dwr-user] memory and performance fixes for 2.0.4 I'll have a poke around and see what I can find about onunload handlers. It occurs to me that one solution of firing a sync XHR and then canceling it after 30ms, could actually work. Joe. On Thu, May 15, 2008 at 11:22 PM, Mike Wilson public.gmane.org> wrote: Joe wrote: We'll need to think a bit about how we handle delete-script-session-on-window-close feature given your fix to the hanging TCP connection issue. I wonder if we could use iframe or similar.. Yes it is a bit tricky as the request by definition needs to be sent just as the page is unloading, and it may as well be canceled before making it "out" on the wire. There is a high probability for different behaviour in different browses as we have seen before. We know that IE's iframes continue executing after page unload but what about the other browsers? (And I guess synchronous XHR is out? :-O ;-) Best regards Mike

Re: Help needed

Joe Walker — 2008-05-16T16:35:16

I think you are probably using an old version of DWR - it would probably be a good idea to upgrade to 2.0.3, and try again then. Joe. On Wed, Apr 30, 2008 at 2:47 AM, Omesh Jadia public.gmane.org> wrote:

Re: how to make the server aware of ReverseAjax

Joe Walker — 2008-05-16T16:24:00

(I've been away at JavaOne, and just catching up) You can do this: import org.directwebremoting.proxy.dwr.Engine; Session s = // the script session(s) of the browsers to toggle active reverse ajax Engine engine = new Engine(s); s.setActiveReverseAjax(true); Joe. On Thu, May 8, 2008 at 10:34 AM, Nitin Thakur public.gmane.org> wrote:

Re: Parameter marshalling between iframes + DWR

Joe Walker — 2008-05-16T16:18:19

Yes please, Mike. Thanks, Joe. On Wed, May 14, 2008 at 4:58 PM, Mike Wilson public.gmane.org> wrote:

Re: memory and performance fixes for 2.0.4

Joe Walker — 2008-05-16T16:17:28

I'll have a poke around and see what I can find about onunload handlers. It occurs to me that one solution of firing a sync XHR and then canceling it after 30ms, could actually work. Joe. On Thu, May 15, 2008 at 11:22 PM, Mike Wilson public.gmane.org> wrote:

Serializing closures

Joe Walker — 2008-05-16T16:15:02

In Javascript: function delayedUpdater(message) { dwr.util.setValue("id", message); } dwr.engine.setActiveReverseAjax(true); Remote.method(delayedUpdater, 4000 /* No callback */); And on the server: public class Remote { public void method(JavascriptFunction delayedUpdater, int delay) { // Fire off a new thread to do some work Thread work = new Thread() { public void run() { // Loop forever pinging the client int i = 0; while (true) { Thread.currentThread().sleep(delay); delayedUpdater.execute("Hello, World: " + (i++)); } } }.start(); } } In short: how about we create a new type: org.directwebremoting.io.JavascriptFunction, with associated converter logic so we can pass a function to the server for execution. I was just doing some work on the dwr data store when I found myself needing something like this, and it seems like it could be generally useful. Thoughts? Joe.

Re: Using Interface as method parameter

Joe Walker — 2008-05-16T15:59:45

Also backlogged, but also interested. Joe. On Fri, May 16, 2008 at 2:43 PM, Randy Jones public.gmane.org> wrote:

RE: Using Interface as method parameter

Randy Jones — 2008-05-16T13:43:52

I'm definitely in. I haven't got to this yet, but still very high on my list. Randy From: Mike Wilson [mailto:mikewse-PkbjNfxxIARBDgjK7y7TUQ< at >public.gmane.org] Sent: Thursday, May 15, 2008 6:12 PM To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter [going through my backlog, sorry for the delay] Hi Randy, I have collected some different requests on the class-mapping stuff, and your dilemma overlaps partly with another request from some time ago. Maybe we should have a discussion about these with Joe and maybe Jose? If either of you are interested in contributing to this stuff then we could open up a design discussion in a new mail thread? Best regards Mike ________________________________ From: Randy Jones [mailto:randy_jones-v48TKWNzwFw< at >public.gmane.org] Sent: den 9 april 2008 15:08 To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter Mike, The JS library I'm using is build on top of dojo so it is using all the inheritance infrastructure dojo provides. My problem is I have a situation where I want to pass an object to the server, via DWR of course, to then do some processing. An example that very well explains my situation, or so I hope. My JS library has an interface called Shape and Rectangle and Circle are concrete classes that inherit from Shape. I would like to have my Java method accept Shape instead of Rectangle or Circle. Originally when I started this project I tried to just pass the object, but I had bad luck b/c DWR marshaled a lot of extra information (handlers etc). So I then went to essentially cloning them using the JS constructors as you mentioned. This was all fine when I was only really working with Rectangles, but not they want Circles too (insert normal developer frustration based on floating requirements). So the JS library has toJSON() functions on all the objects so it's easy for me to just the value representation of an object (no handlers or internal vars) which would then be a great object for me to pass to DWR with the exception of this interface issue. I haven't looked at this part of DWR really, if you think you have a way around this or think this might be a good idea to look at implementing, if you point me in the right direction I'll see what I can work up. Randy From: Mike Wilson [mailto:mikewse-PkbjNfxxIARBDgjK7y7TUQ< at >public.gmane.org] Sent: Wednesday, April 09, 2008 3:47 AM To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter Hi Randy, No, as the class mapping system uses the JavaScript class, objects need to be created with "new " to participate in class mapping. Note though that DWR already does this with outbound objects so all objects you receive from the server will already have the correct class association. How are your client objects created as you hint that you have no control over this? If you have no control over object creation I guess there is only one concrete class to map against, and then maybe you could use the concrete class (instead of interface) as argument in your Java API? Best regards Mike ________________________________ From: Randy Jones [mailto:randy_jones-v48TKWNzwFw< at >public.gmane.org] Sent: den 8 april 2008 22:11 To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter Mike, I have that working for a single case, as your example describes, but I'm looking to expand, so is there also a way of defining what class it is w/o using the JS Constructor? I'd prefer this b/c basically what I'm doing right now is similar to Joe's work with GI where I'm creating Java classes that represent a JS library so you can see I'd rather not have to basically clone the existing objects. To continue with you're example something like: myObj.className = "testdwrclasses.Concrete" I thought I remembered a discussion on something like this but I couldn't find it in the archives, thus prompting my inquiry for documentation I couldn't discover. Randy From: Mike Wilson [mailto:mikewse-PkbjNfxxIARBDgjK7y7TUQ< at >public.gmane.org] Sent: Tuesday, April 08, 2008 5:31 AM To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: RE: [dwr-user] Using Interface as method parameter Randy, I'm sorry, I know that this should have been done a long time ago and it is on my to-do list. Though, like most people I need to do paid work at day-time, and all my free time has been devoted to DWR memory and performance fixes lately. Anyway, this should get you going: 1.Make sure you have converters for both the interface and applicable concrete classes. 2.Add a "javascript" attribute to these converters. Example (dwr.xml): You may then also create new concrete objects from the client through generated JavaScript constructors: var c = new Concrete; Best regards Mike ________________________________ From: Randy Jones [mailto:randy_jones-v48TKWNzwFw< at >public.gmane.org] Sent: den 4 april 2008 23:48 To: users-EyPigyGktj4FDOXUYO6UHQ< at >public.gmane.org Subject: [dwr-user] Using Interface as method parameter Mike/Joe, I actually have a need to use an interface in a method parameter and couldn't find documentation on it, is it somewhere on the site? Randy

RE: Firefox 2.0.0.14 dwr utils bug

wessie — 2008-05-16T12:07:20

Yes, sorry about that one mikewse, I got so caught up with using your original suggestion and then copied my code into the 'test' app and forgot about that button there %-| Sorry making a few amateurish mistakes. Hey, thanks again for all the help! mikewse wrote:

RE: Firefox 2.0.0.14 dwr utils bug

Mike Wilson — 2008-05-16T11:43:20

You could also phrase that as despite not calling any DWR util function the bug still remains. I suggest you isolate your JavaScript<->Firefox bug in an example without DWR and then ask for help in a new post titled "layout problems when manipulating TABLE in Firefox", possibly in a mailing list or forum specializing on JavaScript rendering in browsers. Mike

RE: Firefox 2.0.0.14 dwr utils bug

Mike Wilson — 2008-05-16T11:33:03

Oh, it's nice uploading the file but I am concerned about having the code being discussed available in the mailing list for a long time. Are Nabble uploads "eternal"? I was assuming that they may remove them after some time. I should click on the (broken) image to trigger the bug, and not the button labeled "Trigger" ?! That's not really making it evident for us :-/ I can see your problem now. I'll take a look. Mike

Re: Using Interface as method parameter

Jose Noheda — 2008-05-16T09:57:07

Hi, Fine with me. I'm interested in the topic Regards On Fri, May 16, 2008 at 12:12 AM, Mike Wilson public.gmane.org> wrote:

RE: Firefox 2.0.0.14 dwr utils bug

wessie — 2008-05-16T09:01:29

Humblest apologies mikewse, I attached the html file and then tested the link to see if the 'error' was replicated and i still got the erroneous behaviour and assumed that others were getting the same problem. I also assumed it would be better to upload the file as opposed to sending the application. Sorry for that assumption as well. When i checked the uploaded html i noticed that despite not returning any data from my ajax service (to be expected) in firefox the 'gap' still grew every time i clicked on the image. I'll try to strip down the application and send a working example asap! sorry for the mistakes and thanks again mikewse. mikewse wrote:

RE: Firefox 2.0.0.14 dwr utils bug

Mike Wilson — 2008-05-16T08:50:32

In the file version I can download from Nabble there are still external references in the file, and the Trigger button calls a non-existing function. There is also a lot of JS code that I can't judge if it is important for the issue reproduction. With respect for the people helping you for free, to help you with your paid job on this mailing list, please put effort into making a *minimal* and working example that reproduces your problem. Also, if you are mailing through Nabble please consider inlining the HTML example instead of attaching. Mike