The other question: Is there a way to not to write custom FieldCustomizer
for columns? Can't GWT know which proxy/property I am editing? Or am I doing
something wrong? Thanks!
Assuming by FieldCustomizer you mean FieldUpdater... It seems not – the
Column.getValue method must be
On an unrelated note, the wysiwyg formatting options and copy/paste in the
new google groups seem to be essentially useless, as indicated by the
terrible display of code above... Apologies for sharing that terrible
looking code segment..
-C
--
You received this message because you are
I don't fully understand the interactions between RequestFactory and the
editor system, but it seems that there is another issue which is still
causing this error to occur, though it seems to be another, unrelated bug,
which only affects the HasDataEditor adapter and the editable HasData
From the little information here, it seems that either Child2.html has a
script tag loading Child1.nocache.js or Parent.nocache.js
Parent serves no purpose in your setup - if it were compiled and run, it
would try to start both applications at once.
--
You received this message because you
[ERROR] Hint: Previous compiler errors may have made this
type unavailable
Which type? Which error? If you are missing code in your Child1/2 projects
that disappears only when the Parent module is missing, something is not set
up correctly, or perhaps you are still attempting to compile the
On 2 Feb., 00:27, Colin Alworth niloc...@gmail.com wrote:
[ERROR] Hint: Previous compiler errors may have made this
type unavailable
Which type? Which error? If you are missing code in your Child1/2
projects
that disappears only when the Parent module is missing, something
In your *.nocache.js content, you say you will have 'this.resourceUrl =
content-replaced-by-velocity;' – what is 'this' in this context? If it is
the shared window that all the portlets use, they will each overwrite the
resourceUrl property. If each portlet will use a different module, you
I did, in order to verify the fix, and have just filed an issue.
http://code.google.com/p/google-web-toolkit/issues/detail?id=5981
The patch is below - it can be applied to 2.1.1 or 2.2.0-m1
Index:
user/src/com/google/gwt/editor/client/impl/AbstractEditorDelegate.java
This issues was worked out in ##gwt - had to do with a class that was not
available to the GWT compiler. Still, the error reported is less than
helpful.
--
You received this message because you are subscribed to the Google Groups
Google Web Toolkit group.
To post to this group, send email to
Every Async impl on the client gets a serializer instance to go with it -
crack open the generated code to see what it comes up with. It is possible
to subclass the ServiceInterfaceProxyGenerator to provide your own custom
subclass of ProxyCreator which instead of setting the impl's superclass
What about getting the editor out of the EditorError by calling
EditorError.getEditor()? I know it doesn't get a widget per se, but there
seems to be a good reason for that - not all editors are also widgets, so
you may need to deal differently with each type of editor's error.
My
For my part, I've not been creating a 'createEditorDriver()' method, but
instead 'getEditorDriver()'. This way, there is no concern about binding and
traversing the tree more than once.
Am I in error with this approach? Each time my presenters get a ref to the
driver and have finished loading
Every editor knows the primitive fields it's editing. If the editor
implements HadEditorErrors.showErrors(), it can simply compare
primitiveField.asEditor() with error.getEditor() and highlight the erroneous
primitiveField.
I've read through this at least half a dozen times, but I am
Take a look at the new google groups as an example of what you are saying
:). The default Linker (IFrameLinker) sets things up to start with a js
file, then load the strongly named html file into an iframe. Additional
linkers include building for running as a google gadget, and you can define
The parameter to set is part of the response from the server - the
content-type header can have an attachment property which indicates the name
of the file to be used.
--
You received this message because you are subscribed to the Google Groups
Google Web Toolkit group.
To post to this group,
In the past, I have used DynaTrace Ajax
Editionhttp://ajax.dynatrace.com/ajax/en/Default.aspx,
a free IE profiling tool. It has the downside from hitting you with massive
information overload, but I have been able to use it to find array copying
and dom manipulation which, when modified, was
I don't think you want to directly gzip the files, as some browser don't
support downloading this. Instead, I would suggest gzipping content on the
fly as needed - it is up to you to decide what should and should not be
gziped (note, for example, that RPC calls over a certain size are already
Due to how Java does generics by erasure, this is not possible. Instead, try
the way that GWT.create() works, where a instance of the expected class is
passed in.
For example, your method signature could be
public static final T void callJSONRPCService(ClassT clazz,
AsyncCallbackT callback);
Without more information, I would be hesitant to say you've found a bug. The
RPC mechanism must be as conservative as possible in its estimates of what
can and cannot be serializable, otherwise it would attempt to allow every
widget to go across the wire.
A few things to check – does the type
Assuming MapWidget extends IsWidget, yes, that should work.
On Sat, Feb 19, 2011 at 8:26 AM, pete superp...@geekcity.de wrote:
But I should be able to use an Interface in my template, if I use
(provided = true)?
On Feb 18, 5:22 pm, Y2i yur...@gmail.com wrote:
Oops, sorry I didn't notice
I think you are addressing the wrong issue – Scott is pointing out that
general exceptions do not allow you to re-fire contexts after modifying the
proxies further. The r/o proxy instance is stuck as read only because there
still exists a context-specific edited copy of it, which did not
Can you share a little more on what exactly is happening? The error
message/stack trace (so as to see exactly what is complaining), the
inheritance of your MapWidget interface, and maybe how it should be bound to
a real class (through a replace-with or generate-with, or through a line or
two
FWIW I've seen this happen a variable will always blow an exception (NPE in
my case). At the time, it appeared that the compiler was deciding that it
wasn't worth it to call any of the code, so no 'error happened!' or 'test
finished' message ever occurred at all, and from your example, I think
My approach has been to use RequestFactory.getProxyId to turn known ids into
EntityProxyIds, and from there to get the actual instance. This works for
the most part, unless you actually want to use the same id format that the
server uses. In this case, why not add a method to your
My generators tend to be very heavily commented, and I try to keep blocks of
printed code as small as possible in my code. I like to work up most of the
boilerplate stuff in an existing abstract class or classes, and extend those
when making my generated class. Small methods help too - your
I've already got a generator https://github.com/niloc132/celltable-toolsthat
builds FieldUpdater instances (as well as Column instances, complete
with getValue calls), so it is just a matter of extending it a little
further to make this PendingChange-like thing.
And, of course, to either stop
It should be trivial to make your own call in jsni to $wnd.open, returning
whatever handle you want, or doing the focus right away. Look at how
Window.open is implemented, and see if you can replicate it in such a way to
have the exact behavior you need.
--
You received this message because
This isn't a bug, this is just how the class is defined - DateBox is not a
ValueBox, so attempting to pass it into a function that expects a ValueBox
will naturally fail.
A few options you have: Try making a ValueBoxBase instance that wraps a
DateBox - this may or may not be possible, but
is that the table is
created but no data is added, i mean, no row is created. Do you have any
clue?
Thanks,
Aldo
On Fri, Mar 4, 2011 at 2:54 PM, Colin Alworth niloc...@gmail.com wrote:
I've already got a
generatorhttps://github.com/niloc132/celltable-toolsthat builds
FieldUpdater instances
My understanding is that RF assumes you will be using the same persistence
session for the duration of the request to prevent any perf issues here.
Those find operations are done because RF considers the ID to be the only
valid way to know that it has the latest copy of the object - every time
If one of your editors needs access to the original object, consider
implementing a ValueAwareEditor so it has access to the original model. This
way you can flush changes back as part of the regular editor system.
If you want to make other arbitrary edits outside of the editor system, then
it
com.google.gwt.http.client.URL
The client package in that indicates that this class is only to be used in
client code, code that is compiled to JS. This is causing problems for you,
as part of that class is implemented using native JS, which of course JBoss
can't run.
You'll need to find
The compiler will make the methods it can into static calls. As far as
making an instance static (i.e. making a singleton), this probably won't
gain you anything for a Cell, which has very little state, but in the case
of very large widgets/composites, it might make sense to do.
So Cell,
Couple of thoughts for you to consider:
- If the servlet is building possibly unsafe html, this should be
considered a bug - content encoding should be made safe at the time when it
is outputted, to whatever format is expected on the client. If you were to
send xml or json with or
In general the compiler will select the most specific type it can - if you
write
MapString, Panel map = new HashMapString, Panel();
it will change that to
HashMapString, Panel map = new HashMapString, Panel();
as it is clear that the object can only be a HashMap at that location. In
other cases
On Thursday, August 25, 2011 3:09:22 PM UTC-5, Ryan wrote:
However, if I declare AException in the client, but throw either of
the two child classes in the server, GWT wraps it in an
InvocationException.
This is the key to your issue - if the client can't de-serialize it (because
the
I've gotten emma to work with gwt:test, but only with maven 2 - I havent
tracked down the specifics of why it didn't work with 3. Main piece in doing
this was to stop surefire from running, and make sure that gwt tests run
normally during a build. After that, the only step was to run the
The @ClassName annotation can be used if you have class names in your css
file that are not legal Java identifiers
from
http://code.google.com/webtoolkit/doc/latest/DevGuideClientBundle.html#CssResourceCookbook
--
You received this message because you are subscribed to the Google Groups
NodeList, the return value of getElementsByTagName, has a method called
getLength(). The standard way of iterating through the contents would
involve using a for loop, and testing that n never reaches getLength()
--
You received this message because you are subscribed to the Google Groups
Yes, your JSON is expected to have an object, not a collection, at its root.
This turns out to be a good idea in general as well - in some browsers it is
possible to load json that starts in an array from another site, and read
the contents of those objects, something which is not possible if
Quoting from http://www.sitepen.com/blog/2008/09/25/security-in-ajax/
It is only possible to hijack JSON data with a root that is an array.
When the root is a primitive, primitive values do not trigger a
constructor. When the root is an object, it is not valid JavaScript
syntax, and therefore
Compiling in PRETTY also turns off some of the optimizations, so turn them
back up again using -optimize 9 in your args. This will hopefully allow you
to reproduce the issue, but still see what the code looks like to find
where the error is taking place.
See
GwtTestCase has support for waiting until an asynchronous part of the test
is complete - check out the delayTestFinish(int) and finishTest() methods.
Beyond that, there really is no way to generally pause execution. You can
use one timer to watch another timer, and call cancel on it if it runs
You can't replace the static method through replace-with, but you can make
the static methods call GWT.create on a class that has one implementation,
and use replace-with to swap in another one. For example,
om.mycompany.client.ui.CssGradientUtil.getLinearGradient could be a static
method that
Long values cannot be sent as a number over JSON, because JavaScript only
support smaller numbers. If you only need Double or Integer precision, then
declare that in your autobean, and you can be sure that JSON will pass
those values correctly. GWT automatically allows you to create values as
I'm fairly certain that the compiler will figure out that those methods
return constants and will either optimize them out, or at least turn them
into static methods anyway, so a singleton isn't going to buy you too much.
If you like the code style, that's one thing, but at least in other cases
I thought you were suggesting the static/singleton part here by way of
suggesting that this was a better way of doing things, avoiding
constructing the same instance over and over, but in retrospect I might
have misunderstood. If you are suggesting the static/singleton stuff to
make it easier
You are on the right track for this, but you need to wrap each model with
an autobean: encodeBean.as() is not the same instance as t1, even though it
is obtained from the autobean that wraps t1. You need to wrap each
instance, and include the wrapped instance in the eventual tree to be
The rebinding is how all of the GWT.create calls work, to build browser
specific implementations of most of those (mostly to deal with browser
differences), and to generate the needed source for a few (mostly
i18n/clientbundle interfaces and RPC). This is perfectly normal, and
expected.
As
You are running into a Chome/Dev mode bug that has apparently been around
since about Chrome 10. It seems to be intermittent, only happens when
stepping in and out of JSNI methods, and will not affect you application
once it is compiled.
More information, discussion:
It could be possible to wrap your ClientBundles in an appearance
implementation, and use replace-with declarations on that, to check for dpi
when the app starts up. Check out the notes on the appearance concept at
You could wrap them up as a TextResource in your ClientBundle and inject
them into the page, but Android 1-3's Browser doesn't support SVG, nor do
IE versions prior to 9.
On Thursday, April 19, 2012 12:38:54 PM UTC-5, Evan Ruff wrote:
Joe,
SVG would be awesome if my sources were vectors. By
, etc?
Thanks,
E
On Tuesday, April 17, 2012 4:40:35 PM UTC-4, Colin Alworth wrote:
It could be possible to wrap your ClientBundles in an appearance
implementation, and use replace-with declarations on that, to check for dpi
when the app starts up. Check out the notes on the appearance
With soft-permutations though, this could cut back on the explosion, though
you are right that the app would be unable to switch at runtime between
which version it is running. That said, once a user has been set to a
particular set of features, you probably want to keep them there. And a
The callbacks you are configuring aren't running until the later test is in
the middle of working. You are using the finishTest() and
delayTestFinish(int) correctly in your first test, but the second has a
pair of asserts going off asynchronously, but you are not delaying that
test's
GWT 2.4 included a custom copy of htmlunit, apparently built from rev
5940[1] of the htmlunit's source - the following unit test passes under
that version using the com.google.gwt.xml.XML module:
public void testSelectElement() {
String xml = rootchild/childchild /child
, it'll only be run in real browsers (when using -runStyle Manual,
RemoteWeb, Selenium or ExternalBrowser)
On Tuesday, July 17, 2012 10:34:31 PM UTC+2, Colin Alworth wrote:
GWT 2.4 included a custom copy of htmlunit, apparently built from rev
5940[1] of the htmlunit's source - the following unit
All of the source for the existing RPC proxy generation code can be found
in gwt-user.jar. As with all Generators, the first step is to create a
Generator or GeneratorExt subclass, and reference it in the module (this is
in RemoteService.gwt.xml):
generate-with
Your catch (Exception e) should also be printing out the results, but your
quoted error message didn't contain that exact json - any chance that the
json you started with isn't actually making it to the client? One simple
way to test would be to escape that string and put it in your Java code
The DOM class is used to bridge the various gaps between browsers - the
dispatchEvent method, as you've noted, is invoked by the specific
implementation in use at a given time. These browser-specific
implementations live in the com.google.gwt.user.client.impl package:
*DOMImpl
For what it is worth, Maven 2.2.1 with the stock Emma plugin (version
1.0-alpha-3) works just fine with gwt-maven-plugin:test. There are
certainly issues with Maven 3, but as we run test coverage outside of our
normal nightly builds, we drop to Maven 2 for coverage details.
With the gwt:test
There is one known issue that is affecting some users who update from GWT
2.4.0 to GWT 2.5.0 detailed in
https://groups.google.com/d/topic/google-web-toolkit/H0Blv3C_fns/discussion.
The easiest workaround seems to be adding requestfactory-server.jar to the
server's classpath, but anything
TimeField, along with all other ComboBox subclasses, by default filters
results in the drop down based on what text is present in the text field -
thus, if only one value matches, that is all that will be displayed. To
turn this behavior off, invoke the setTriggerAction(TriggerAction) method
In addition to matthew's comment, you are invoking a method that apparently
has two arguments
callFacebookAPI(Ljava/lang/String;Ljava/lang/String;)
with only one:
(facebookUrl)
Delete one of the two Ljava/lang/String; parts on the method invocation so
you actually point to your method
I've not seen this specific issue (GWT 2.5.0 from Juno or maven command
line on both mac and linux) in my projects that use RequestFactory, at
least not from a source where I can point to as our own JSNI mistakes.
One kind of error I've seen that only emerges in OBF is JSNI local variable
thats baffling me is that this is all code that has worked and
worked with this version of the tools. I must have changed something small
that has blown things up.
On Thursday, December 13, 2012 11:33:52 AM UTC-8, Colin Alworth wrote:
I've not seen this specific issue (GWT 2.5.0 from Juno
It looks like it thinks you are sending the CustomFieldSerializer itself
over the wire, so wants to find a serializer serializer... The extra
prefixed package name com.google.gwt.user.client.rpc.core indicates that
it wasn't able to find your field serializer in the normal package where it
Greetings from IRC - glad to see you are continuing with this idea.
It could be worth considering inlining the sizzle.js code as JSNI to avoid
the extra .js file at all, either from the module file or from the base
html file. This has the additional constraint/advantage that you'll need to
Also GWT 2.4, and is it possible you have more than just 2.4 on the
classpath?
Do you have the GWT-Incubator on your classpath? I've seen cases where the
incubator's extremely outdated ClientBundle generators try to take over and
end up just breaking things.
Otherwise, can you share the full
Think about this question from the other way around - same interfaces, but
now look at the data itself, the JSON:
{name:foo, shape:bar}
What is that? Is it a Binterface with its other properties null? Is it a
Cinterface with its other properties null?
{name:foo, radius:100, height:200}
What
The skype browser plugin adds additional style tags to the page, and mess
up StyleInjectorImplIE's assumption that no one else is spitting out style
elements. SnagIt used to (and may still) cause exceptions in various IE
versions when some elements resize/reflow (I never nailed down exactly
What about the first could be optimized out - just the size() accessor?
Most List implementations have a size field (see
http://code.google.com/p/google-web-toolkit/source/browse/trunk/user/super/com/google/gwt/emul/java/util/ArrayList.java?r=6609#77
for the actual ArrayList used in compiled
JavaScript doesn't have a concept of type, so all type checks are really
just function calls. Both instanceof and casts need to be generated to have
JavaScript behave exactly like Java. Here's a quick demonstration:
public class Test implements EntryPoint {
public static class SomeObject {
I can definitely confirm that with the closure compiler enabled the
compiled size drops on the few apps I've tried it on, on the order of 5-15%
(no hard and fast numbers yet, working on such a writeup now). I can
confirm both a performance and size improvement with turning off cast
checking, but I
We've been using the Maven2 emma:emma goal, with no modifications at all -
seems to behave correctly with htmlunit in dev mode, both for traditional
junit tests and GWTTestCases. No changes to the pom, just executing
emma:emma with maven 2.2.1 on our project.
On Thursday, March 7, 2013
:13:11 PM UTC-5, Colin Alworth wrote:
We've been using the Maven2 emma:emma goal, with no modifications at all
- seems to behave correctly with htmlunit in dev mode, both for traditional
junit tests and GWTTestCases. No changes to the pom, just executing
emma:emma with maven 2.2.1 on our
In most cases that is true, but CssResource is an interesting exception.
The fallback rules work by looking to see if there is no implementation for
a particular value, and if not, looking to see if there *is* an
implementation for some other value. In the case of ClientBundle and
Without specifics of what may be referencing those objects it is hard to
say for sure, but memory details while running in dev mode will likely not
reflect the compiled code. To properly test your memory patterns, compile
to JavaScript with style PRETTY or DETAILED so you can read the code, and
If you compile in PRETTY instead of DETAILED, it won't intern those
strings, but still will leave the output mostly readable (just no packages).
Without seeing the rest of the structure of the module files, it is hard to
speculate, but we're using more or less the same idea successfully, though
We've found experimentally that the meta tag has no effect on IE8 when in
intranet mode. We've further found that it does seem to respect the http
header, which could be set in a filter like this:
public class LatestIEFilter implements Filter {
@Override
public void doFilter(ServletRequest
=false /
I'll try to build a more complex test case that actually uses the
properties defined, just in case the compiler is getting to clever for me.
On Monday, October 28, 2013 12:45:45 PM UTC-5, Colin Alworth wrote:
If you compile in PRETTY instead of DETAILED, it won't intern those
strings
Chak, take a look again at my post - while the meta tag definitely does not
work to tell IE8 to behave when in intranet mode, loading the exact same
html content and sending the same ua-compat details over a HTTP header
*does* solve this.
On Monday, October 28, 2013 3:08:47 PM UTC-5, Chak Lai
Line 155:
ssize_t n = recv(sock, readBuf, BUF_SIZE, 0);
That seems to say 'block until bytes are written by the other end of the
socket', or in other words, wait until the JVM is ready to go. Are you sure
that the IDE isn't paused on a breakpoint, or that you have waited long
enough for the
suggestion and check now the other side, but it's good to
know that you haven't had an issue with FF and Mavericks as then that would
sadly imply that there's something wrong with my computer :(.
On Friday, November 8, 2013 9:20:10 PM UTC-5, Colin Alworth wrote:
Line 155:
ssize_t n = recv(sock
By default, Selenium starts the browser with a fresh profile every time,
which means that it has no plugins installed. On quitting, it deletes that
profile again, to make sure that it won't slowly consume your disk.
When you start the firefox driver instance, you can ask it to load profile
None of the WEB-INF/ directory should be needed, provided you are not
running a servlet container, but that is the only 'server-only' code that
generally is created that will take up any meaningful size.
Another option at your disposal is to get rid of the permutations and
instead make just
Another new toy to propose for saner Selenium testing:
https://github.com/niloc132/gwt-driver
On Monday, November 11, 2013 8:59:49 AM UTC-6, Ed wrote:
If you search the forum well, you will find all you need.. Example:
We're using this at Sencha with pretty good success, though it really is
designed to be for testing applications, not libraries. I know of a few GXT
customers who are using it with gxt-driver for widget support, but it is
still pretty early - either everyone thinks it works out great, or there
to the
plugin, I then try it in root mode and works...
The devmode process and eclipse are running with my account, so I can't
tell what else I could have messed up but the important is that I can work
and move on.
I appreciate the help,
On Saturday, November 9, 2013 3:10:58 PM UTC-5, Colin Alworth
Can you try compiling in PRETTY? This will make the retaining tree graph in
chrome's inspector easier to tell where things are coming from and what is
tracking them. But at a glance, something registered something with
google.visualization, and didn't unregister it. Knowing what gD, iv, and fv
Good to know - that makes it seem rather likely that this leak is entirely
within the core visualization code, rather than anything specific to gwt. I
would look at plain JS examples of how to use the library, and make sure
that a step isn't being missed either by virtue of writing in GWT or by
Can you confirm that you are hitting the Compile button in each browser and
that the SDM console is indicating that it is recompiling for each other
user agent? It sounds as though you might be compiling when you start up
one browser, then just turning dev mode on without recompiling in other
There have been a few quick discussions about this in ##gwt on irc, and
while I think we probably need to both move this to -contrib before much
longer but also get some of the compiler experts involved, I suspect this
is going to end up being a specific tool to optimize sufficiently c-like
would expect that site would be working with the version of software
currently deployed for all browsers and not only chrome.
Right now I am getting blank pages in firefox and IE.
Vassilis
On 02/08/14 03:07, Colin Alworth wrote:
Can you confirm that you are hitting the Compile button
Jens is dead on - several API changes from 2.4/2.5 to 2.6 make it difficult
for a library to stradle that divide. We'll be shipping a GXT 3.1 beta Real
Soon Now to enable users to switch to GWT 2.6. Breaking changes include:
- Changing permutations (ie6 and opera are gone, ie10 was added,
There is a prototype project enabling Eclipse to debug the JS running in
the browser with sourcemaps - check it out at http://github.com/sdbg/sdbg.
On Thursday, November 15, 2012 8:46:26 AM UTC-8, Clint Gilbert wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
If I could hook Eclipse
Thanks for posting about this Oskar. The radian/degree code can likely be
changed (is that behavior specified as part of the Java spec?), though
changing RPC serialization sounds like it would come with performance
penalties - any comments there?
What is your preferred method of resolving the
You actually have three properties, not two - the 'extend-property'
declaration means 'add these to the existing list of values'. Running the
SOYC report will give you the full story on your properties and
permutations (and lots more besides).
Try a set-property line to specify not only 'add
Looks like codehaus's snapshot repo is down, so I'm unable to get the
latest gwt-maven-plugin. This url gives a 504 gateway timeout from ngnix
for me:
https://nexus.codehaus.org/content/groups/snapshots-group/org/codehaus/mojo/gwt-maven-plugin/2.7.0-SNAPSHOT/maven-metadata.xml
If I set gwt-user
It sounds like you have non-gwt-capable classes in packages meant for GWT -
is that deliberate? For example, test classes to make sure the various
server components in your project work, but they are in your .client or
.shared package?
If they are not, then GWT will totally ignore them, as no
1 - 100 of 463 matches
Mail list logo