First start SDM, then deploy your war to your external server. SDM
generates a special *.nocache.js file that enables automatic recompilation.
You have to make sure that this special version is deployed.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
I think your Settings class should have @JsType(isNative = true, namespace
= GLOBAL, name = "Object") to make it a JavaScript object literal.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To unsubscribe from this group and stop
Just use CssResource.
http://www.gwtproject.org/articles/gss_migration.html
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to
Well then you have a class path issue. The classpath of your CodeServer run
configuration is probably different than the one used for a normal compile.
Make sure you do not have GWT 2.7 on classpath.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Yes it can happen that SDM works but a production compile fails. For me its
pretty rare but it can happen. I guess there are some difficult to spot
bugs within incremental compilation of SDM.
You can clean SDM caches by launching SDM, open its site (localhost:9876)
and hit the small clean
> GWT is on 1.0.0.1.GA. Isn't that going to be mostly compatible with
> 1.1.0? It seems like this would just require changing the version in a
> pom, and the rest would work. I would be really surprised if there are any
> breaking changes between these versions.
>
Upgrading Validation API
I think that should just work because @JsFunction is usually used to
provide callback functions to JavaScript. I don't know your exact API but
something like
@JsFunction
interface Action {
void exec();
}
@JsType(native = true.)
class Item {
// Alternatively use java.lang.Runnable in
> Once CodeServer.main() has been run, is there a way to stop the code
> server, such as in the tear down method of the test?
>
Hmm no. I guess you would need to run it as a separate process and then
kill the process. But always starting/killing SDM also costs you time so
personally I would
> How can that be done? Can anyone point me to which code class / method
> needs to be called for starting up SDM?
>
CodeServer.main()
> Also, if there's a better way of doing what I'm trying to do, or if anyone
> has done this before, please share your thoughts.
>
IMHO you would usually
Its not an UiBinder construct because within a HTMLPanel (which the example
uses as root) everything is treated as HTML unless its a UiBinder tag. So
what you are seeing in the example is plain HTML with Polymer extensions
like the "is" attribute.
You can read that up
at:
> On Friday, April 1, 2016 at 1:55:51 PM UTC+2, Jens wrote:
>>
>> Sounds like you are not the only one:
>> https://groups.google.com/forum/#!topic/google-gin/ywh7d9s6GpA
>>
>
> Looking quickly at the stacktrace, it could be a class from super-source:
> no
> Is the CompletableFuture on the emulation implementation plan for 2.8
> release? It's an amazing helper for chaining asynchronous operations.
>
I think the consensus between Colin, Andrei and me (we have done most of
the Java8 API work so far) is that we want to emulate CompletableFuture
*___jb_old___ files are temporary files created by IntelliJ when safe write
is enabled (Settings -> Appearance & Behavior -> System Settings -> Use
"safe write").
Looks like the file has been deleted by IntelliJ between the time GWT has
detected it was created (using Java watch service) and
> I'm familliar with the small 'pause' icon. It isn't being triggered. I
> actually think that the error is being thrown before Chrome starts running
> the code. Does that even make any sense?
> In the chrome debugger, if I go to sources, the .js file doesn't
> even show up under the js
Given that DevMode in GWT 2.8 launches SDM by default and SDM uses
obfuscated JS output by default, should we add -style parameter to DevMode
so people can opt-out of obfuscated JS output?
See: https://groups.google.com/d/msg/google-web-toolkit/yfhiStlMW7E/0MUPV1EwBgAJ
-- J.
--
You received
Hm I see. You probably have done DevMode -style, right? However in GWT 2.8
that DevMode class launches a different class called CodeServer which you
can also start yourself. This CodeServer class has a -style parameter that
you can use to modify the output. The corresponding commit to make that
You are aware that GWT compiles Java code to JavaScript code and thus once
you uploaded a java file you would need to compile it to JavaScript in
order to display anything in the browser?
What you want to do would mean that your server needs to modify the source
code of your app and then
It's expected behavior in GWT 2.8. When starting DevMode there should be a
console output saying "Starting CodeServer with parameters " so you can
see how the SDM CodeServer is launched. It should contain -style OBF for
obfuscated JS along with -XmethodNameDisplayMode. I think the primary
You have to go through all *.gwt.xml files of GWT SDK and search for
deferred binding rules that match the user.agent property
against safari / firefox. Also the GWT compiler itself might execute some
visitors specific to a permutation on the parsed AST. These visitors can be
found in
> If your concert is to be able to write one-line initializers, and you are
> using java8 syntax you also do something like that:
>
> public static T apply(T t, Consumer fn) {
> fn.accept(t);
> return t;
> }
>
> @JsType(isNative = true, namespace = JsPackage.GLOBAL, name = "Object")
>
Client side is insecure by definition as you can not control what will be
done with the downloaded Javascript code.
If you have implemented authentication and authorization correctly in your
server side GWT-RPC methods then its not an issue.
-- J.
--
You received this message because you are
With old/classic DevMode -noserver there is no jetty started at all so it
did not matter if you have a jetty-all-9.2.10.jar on classpath or not.
However with GWT 2.8 when executing DevMode -noserver the SuperDevMode
CodeServer will be started which uses its own embedded jetty
server/library.
I guess your border-style is still set to "none" (the default value)
because you have not set any. You need to set it to "solid" for example.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To unsubscribe from this group and stop receiving
If its just for value classes and factories you better use Google's
AutoValue / AutoFactory projects which are based on annotation processing
instead of hacking the Java compiler as Lombok does.
Personally I would never use Lombok. As it adds code during compilation,
that code is invisible to
Ok I have added a bunch of feedback comments directly in the document.
In general I think its fine although we should try hard to minimize labels
and use good colorization of labels.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Contributors" group.
> It would be great if we had an option to tell GWT to not fail on
> compilation errors but automatically mark non compiling code as
> @GwtIncompatible.
> Is it a possible enhancement I can submit?
>
But then you would hide real compilation issues.
> So for now, the only way to use
> Shouldn't the second method be considered as dead code and automatically
> removed by GWT?
>
The following is just speculation as I don't know the compiler internals
out of my head:
I assume thats because GWT will compile the source using JDT in order to
get an AST that GWT can then
> Can I continue the same approach 'converting' each instance to Material
> versions? I'd stil like to use some of my existing styling / css.
>
I never used the library and I don't know you app. So no idea, if
everything that this library offers fits your needs. But I am pretty sure
you can
> My current gwt application does not use UIBinder framework (just
> traditional Java). Do I *have* to switch to UIBinder if I want to use
> GWT Materials?
>
No, of course not.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To
GWT compiler does dead code removal automatically when doing a production
compile. In SuperDevMode no optimizations are done though because
compilation time in SuperDevMode should be as fast as possible.
Take a look at http://gwtcreate.com/2013/videos/#compiler-deep-dive
-- J.
--
You
These xml files are used to generate Java code which then gets compiled to
Javascript and that Javascript is what you deploy. So if you change the xml
you need to rebuild your app.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To
Buts it also not a big deal to have AppProduction.gwt.xml for your normal
production builds and AppUiTesting.gwt.xml which inherits
AppProduction.gwt.xml as well as Debug.gwt.xml.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To
> Yes "processing of 'ensureDebugId()'" I meant, runtime overhead.
> And why do you think it won't require extra 'Javascript' download, as if
> you want to set debugIds you need to inherit Debug module.
>
I just did not mention it because I thought it was a logical consequence
that your JS
> 1- The JavaScript download and processing of 'ensureDebugId()' is an
>> unwanted overhead
>> 2- when we set fixed dubug Id, someone can easily inject Unwanted script,
>> which can make use of this ID. (Not sure who the injection will work
>> though.)
>>
>> Is there any other reason why we
Since "click" is a browser event you would need to use addDomHandler()
instead of addHandler().
If that does not work either then I would let the Cell publish a custom
Event on the EventBus. Maybe using custom events via EventBus is more
readable anyways.
-- J.
--
You received this message
> I have a regular html page. I can use GWT to get one element from that
> regular html page. I want to use GWT to generate a mouse over event on that
> element. How do I do it in GWT code?
>
Never tried it but from the API it looks like to generate a native event
you need to do:
Document
> Unfortunately this leads to massive performance loss since the monolithic
> js is way better and more optimized.
>
I don't think thats true. If you combine all your small apps and the main
app then, yes, the download size will be larger than just one big app
because all your small apps
Never looked at the MobileWebApp example in detail but if the
implementation is too abstract/complex for your own use case then don't use
it.
You are right that the bare minimum would be a class that detects the
orientation change (using various techniques, e.g. resize event or
Read: http://www.draconianoverlord.com/2014/04/01/jdk-compatibility.html
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to
GWT-RPC is just one convenient way to communicate with a server using GWT.
However GWT-RPC uses its own serialization format that is based on JSON.
While convenient to use GWT-RPC also has some pain points that are often
discovered later.
If you just want to make a POST / GET to an URL and
> These exceptions on the next request are problematic for us because that
> means all the data entered into a form and being submitted will be lost
> when the exception occurs. How are you dealing with these kind of issues?
>
The app uses aggressive auto saving when editing. Only issue is
>
> Feel free to get a patch up for review in Gerrit if you have time, I'll
> review it. Question: do we put those in a .github/ folder or directly at
> the repo root (next to the README)?
>
I would put issue/pull request templates into .github/ and a possible
contributing.md on the root.
I think you either have to create a custom JS function that does the
default value thing so you can call that function using JsInterop or you
need to return Double (so the method can return null if devicePixelRatio is
not available) and do the default value thing in Java using an additional
I noticed that the Github issue tracker has a large number of labels and
IMHO most of them are not used anymore. If time permits I thought about
cleaning labels up so its more streamlined:
- Convert Milestone labels to Github Milestones
- Adjust category labels to
match
> Why GWT is designed that way, so hard to extend? Is that a reflection of a
> desperate need for contributors to the project, by having to "fork the
> whole thing"? Or is it to force the developers to use paid frameworks built
> on top of GWT?
> Why package-scoped classes?
> Why private
> Isn't GIN unmaintained anyway?
>
(and what we're all waiting for is proper GWT support in Dagger 2 ;-) )
>
I don't know. GIN works well and a ton of apps use it. A lot of these apps
will probably not migrate to Dagger 2 anytime soon (especially not if it
takes ages for a PR to be
> Hey, thanks for the hint with baconApi but it's the same that the most of
> our users use IE.
>
You could just install a beacon API polyfill that will do a synchronous
XmlHttpRequest, e.g. https://github.com/miguelmota/Navigator.sendBeacon
Then you can transparently use the beacon API
Normally you would add a classname to your widget and then inject a
stylesheet that makes use of that additional class name.
For example if you have a widget whose root CSS class is .myWidget and you
want to make one instance of that widget red then you can add the class
.makeRed and inject
We deploy the new version next to the old version and then use a reverse
proxy in front of the app servers to switch customers separately to the new
version. Once switched to the new version the app gets an exception on the
next server request and that exception tells the GWT app to show a "you
We once did something similar although not on the web but we also had the
issue that users suddenly go away without releasing the lock.
So we gave the lock a lastAccess timestamp and allow others to
override/cleanup the lock if it was older than X minutes. Whenever the lock
owner edited the
> Unfortunately GWT 2.8.0 still does not support that feature. Had to use
> literal("") as well here.
>
You are still using Css right? According
to https://github.com/google/closure-stylesheets/issues/59 closure
stylesheet should support and thus GWT GSS should support it.
-- J.
--
You
You can use any package structure you want. Your server classes can be in
any package but you must make sure that the GWT compiler does not see them.
The GWT compiler only sees code that is included using
in your *.gwt.xml file. By default GWT compiler looks in the package called
"client"
> I'm also interested in this option to reduce the number of permutations
> and therefore speedup compile times (currently around 90 minutes due to
> multiple languages and supported browsers).
>
Code splitting has nothing to do with permutations. If you want to reduce
the amount of
AFAICT you still need JSNI for it. See my (untested) example in
https://groups.google.com/d/msg/google-web-toolkit/YOWINYAoCrI/gFE_KSFcFQAJ
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To unsubscribe from this group and stop receiving
> I have tried stepping through the code in debug mode and found that the
> code indeed updates the TextBox. But the new value just doesn't show up in
> the UI.
>
I am pretty sure the above code works as expected when creating a demo
project. So I guess you have a bug in your code that you
Do you have any specific issues by following whats said in the document:
"You can mark each method in a JSO with @JsOverlay (that’s actually what
they are) and replace the JSNI by following the instructions above."
So basically JSNI methods will become JsInterop native methods if they
directly
> does the class still have to extend JavaScriptObject?
>
No.
--
You received this message because you are subscribed to the Google Groups "GWT
Contributors" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to
Welcome,
for general GWT questions please
use https://groups.google.com/forum/#!forum/google-web-toolkit
This group is for questions about contributing to GWT SDK itself.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Contributors" group.
To
>
>
> .desktopPanel{
> background-image:
> url("\cti\gwtdemo\client\images\intelview-bkground.jpg");
> background-color: lightgray;
>
> }
>
Your image url must point to a valid URL when your app is deployed. I am
pretty sure that images inside your java src folder are not available to
Make both lines the same:
https://gwt.googlesource.com/gwt/+/master/dev/core/src/com/google/gwt/dev/DevModeBase.java#156
https://gwt.googlesource.com/gwt/+/master/dev/codeserver/java/com/google/gwt/dev/codeserver/Options.java#529
One uses getHostName() while the other uses getHostAddress()
--
You are trying to send a collection of java.lang.Object which is not
allowed by GWT-RPC.
See documentation
http://www.gwtproject.org/doc/latest/DevGuideServerCommunication.html#DevGuideSerializableTypes
As you are using JPA you should also read the section about serializing
enhanced classes
> are there any plans/thoughts of using DefinitelyTyped with GWT - maybe in
> GWT 3.x?
> http://definitelytyped.org/
>
> i.e. it would be great to generate GWT js-interop code from the
> DefinitelyTyped type-script definitions.
> then GWT would immediately have type-safe access to all the
>
> AFAIK GWT 2.8 will have jsinterop and elemental so the native
> scrollIntoView() will be used automatically.
>
Only if you use Elemental of course ;)
I don't have a problem to file a bug report in GWT but I thought I should
> ask first here about the reasoning of the current status.
>
> stupid suggestion but do you sink the events you want to listen for?
>
addDomHandler() automatically sinks the corresponding native event type.
You only need to sink events yourself when you override onBrowserEvent() to
handle native events without registering a handler using
Even the old spec wasn't able to do that directly without using a helper
method that maps to a one liner of JSNI code [1]. So with the current 1.0
spec you probably need to do
@JsType(isNative = true)
public class UintArray {
public UintArray(int size) {}
@JsOverlay
public void set(int
> a good choice now could save money, angry peoples and long nights spent
> for nothing! :)
>
A good choice now is to do nothing, because GWT 3 / J2CL is not released
and you don't know if someone (e.g. Vaadin, Sencha, contributors) will
migrate the current Widgets / Elements API to future
> Anything but please not React or Angular :)
>
> I still dont believe this is supposed to be the future on web development.
>
The future of web development does not exist because it changes constantly.
Choose whatever works for you now as long as its backed by some larger
company so chances
Yes sounds like the ErraiBus holds a reference to your class even though
you don't use it anymore. Its similar to GWT EventBus that usually exists
as a singleton in your app and if you don't deregister your handler you
prevent garbage collection.
As Errai is pretty complex and I never used it
Well that behavior is normal when working with floats / doubles, thats not
GWT nor browser related. Computers simply can not represent all floating
point numbers accurately internally. See: http://floating-point-gui.de
So your solutions are:
- Use NumberFormat / Math.round to display the
>
> I get what you say, and agree that its not a very good approach.
> But my concern is, For a huge application having around 200 events . Will
> I have to write 200 Event handlers, 200 Event Classes and Event types ?
>
There is https://github.com/google/gwteventbinder which avoids defining
> Not sure if that property still exists in GWT 2.8 beta as we use some GWT
> snapshot build thats in between 2.7 and 2.8 beta releases.
>
Property still exists, see:
https://gwt.googlesource.com/gwt/+/master/dev/core/src/com/google/gwt/dev/resource/impl/ResourceAccumulator.java
-- J.
--
On Mac OS we use the system property -Dgwt.watchFileChanges=false when
starting SDM. With the property set to false I think it does a file scan
instead of using some watch service. Somehow watching file changes did not
work well on Mac OS.
Not sure if that property still exists in GWT 2.8 beta
>Is the GWT 2.7.0 release note: "GWT Designer doesn’t work with 2.7
> and is no longer supported." accurate?
>
Yes thats correct. No one maintains GWT Designer and thus it has become
incompatible to GWT 2.7+. However the source is now available and you could
patch it if you really
I think Ray Cromwell did something like that using a generator and a
linker. Can't remember exact details and he never shared is experiment. I
would guess it was a resource generator so you can define a
JavaScriptResource in ClientBundle (instead of TextResource) and then this
resource made
> With that being said, I'd really love to reattain the level of
> productivity that I perceive with DevMode when it comes to debugging. SDBG
> just doesn't even come close to what I consider a real debugger. Stepping
> and watching variables is just by far not enough by today's IDE
> Same behaviour with -*draftCompile*, I've also tried *optimizationLevel*
> to 0
>
Hmm and you are double sure that the compiler is really called with -
*generateJsInteropExports* and not just SuperDevMode?
Also in some rare cases the incremental compile cache of SuperDevMode can
cause
I *think* it does not look too bad. You can compile your app using -mode
PRETTY (or DETAILED if you also need package infos) so you can read the
final JS output to see what went wrong. The only difference between
SuperDevMode and a normal compile is that a normal compile optimizes your
code.
Even though Julien said it will be added for GWT 2.8 it is not yet
implemented and there is no patch in review for that feature. So for now
you can not register custom GSS functions without patching GWT.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Not yet supported by Closure Stylesheets,
see: https://github.com/google/closure-stylesheets/issues/16
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to
Basically ScrollPanel has an internal div element (the "scollable" element)
that does not have any height. That is generally fine because the
scrollable area should adapt to the size of its single child but that also
means you can not use "height:100%" for the child (percentages require
height
The browser only offers storing String keys / values in local storage. That
means you need to serialize objects somehow. JavaScriptObject / AutoBeans
can easily be converted to a JSON string. Otherwise you probably need a
library, something like https://github.com/nmorel/gwt-jackson
-- J.
--
Correct GSS syntax is without the dot on the external declaration, e.g.
@external header;
.header {
...
}
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to
> I find it a bit odd. The group reads google-web-toolkit and so as pretty
> much everything associated with GWT.
>
Actually this group is named "GWT Users" (which matches "GWT Contributors",
"GWT Steering"). Only the URL hasn't been changed to not break existing
links to existing
> May be we should consider this in 3.0.
>
There was already some talk about wether or not renaming GWT when 3.0 is
released as 3.0 will likely be very different. But at the end its a
decision made by the steering group. If they are not sure about it they
will probably do a poll and ask the
> Thanks.. Do you know the name of the runner class used to run GwtTestCase
> ?
>
JUnitShell (which is a subclass of DevMode) sets up the test environment
and executes GwtTestCase.
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users" group.
To
Debugging using SuperDevMode is different but definitely not impossible. It
also doesn't cost hours spotting a problem. Eclipse users often use the
Eclipse plugin "SDBG" which allows you to debug your JavaScript code in
your IDE: https://sdbg.github.io . IntelliJ users can do that out of the
Boolean and Double are now unboxed raw types when compiled to JS. I think
that means inside JSNI methods a Double should be treated as double
primitive.
So something like
protected native void setValue(Element e, Double value) /*-{
> There must be a web.xml used by the embedded jetty somewhere while running
> the tests - right ? Is this accessible to us to modify ?
>
There is no web.xml. Embedded Jetty is configured programmatically. To
register a servlet I think you need to use the tag in your
gwt.xml file.
-- J.
> Well in that case, everyone would like to know what the G there stands for
> :) any clues ?
>
>From http://www.gwtproject.org/GWTPolicy.html
*"Name is spelled GWT - all capital and it stands for GWT Web Toolkit."*
-- J.
--
You received this message because you are subscribed to the
> Also, why is the latest jsinterop spec not included in the GWT 2.8 beta
> release?
The JsInterop spec is linked in the release notes:
http://www.gwtproject.org/release-notes.html#Release_Notes_2_8_0_BETA1
No one has written a markdown document for the spec yet, so its not a
dedicated
> Is it intentional/necessary that gwt-servlet.jar is compiled for Java 7
> and not Java 6?
>
gwt-servlet.jar isn't really compiled, its just assembled. The build script
compiles gwt-dev and gwt-user using Java 7 as minimum and then
gwt-servlet.jar is created by pulling in a subset of
> Awesome. Thank you. @JsOverlay did the trick.
>
But IMHO @JsOverlay is only a workaround because it is meant to be used for
code additions that do not exist in the underlying native type. So in your
case GWT will now generate some additional code for your constants and will
not use
> The https://github.com/gwtproject/gwt/issues/9074 was in master. Somebody
> know how to check if it will be included in 2.8 release? the commit is
> https://github.com/gwtproject/gwt/commit/dd215d81ce29db42b185ffa2f86f6af36076f3be
>
Its included in 2.8.0-beta1 (see little tag icon + text on
You can do separate projects just fine if you make your shared modules
separate projects as well so that they become libraries and can be included
in either of the two main projects. No need to duplicate shared stuff.
Alternatively just make one app and
- use some url patterns to make the app
You can use to define a fallback value in case the
user.agent is not recognized. That way you can serve a default permutation
for unknown user.agents. By default the fallback value is "unknown" and
thats why GWT tries to load a permutation file called "unknown.cache.js".
So it should be fine
To debug your server side code you need to start Tomcat in debug mode and
then attach your IDE as a remote debugger:
http://wiki.apache.org/tomcat/FAQ/Developing#Debugging
I assume you are using GWT 2.7 and you are using SuperDevMode. To debug
your client side GWT code you would use your web
I would say: Don't render all available messages at once. Just render the
latest 30 message and if the user wants to see older ones the user has to
tap a "load older" button or you just load them when scroll panel reaches
the near top.
There is rarely a need to render everything up-front
> Any updates about the starting guide of JSInterop?
>
Everything you need to know for GWT 2.8 JsInterop:
https://docs.google.com/document/d/10fmlEYIHcyead_4R1S5wKGs1t2I7Fnp_PaNaa7XTEk0/edit
-- J.
--
You received this message because you are subscribed to the Google Groups "GWT
Users"
> Anyone have a fix yet?
>
Which Gradle version do you use? Gradle didn't support pinning a specific
SNAPSHOT build some time ago. Maybe you should upgrade Gradle.
https://issues.gradle.org/browse/GRADLE-2784
-- J.
--
You received this message because you are subscribed to the Google
> Thanks Lars for the clarification. Looks like a bad plan to fix all broken
> implementations using ObjectInputStream and better remove it completely in
> a GWT-RPC environment.
>
You can use Java serialization just fine as long as you can be sure no one
has modified your serialized data
501 - 600 of 2944 matches
Mail list logo
