[gwt-contrib] Re: Ensure all ProxyAutoBeans not directly referenced from root object or via a shim can be garbage-... (issue1451819)

2011-06-16 Thread t . broyer 


http://gwt-code-reviews.appspot.com/1451819/diff/1/user/src/com/google/gwt/core/client/impl/WeakMapping.java
File user/src/com/google/gwt/core/client/impl/WeakMapping.java (right):

http://gwt-code-reviews.appspot.com/1451819/diff/1/user/src/com/google/gwt/core/client/impl/WeakMapping.java#newcode104
user/src/com/google/gwt/core/client/impl/WeakMapping.java:104: for (;;)
{
On 2011/06/16 00:53:37, rjrjr wrote:

why is this tight loop okay? What makes the thread sleep or whatever?


ReferenceQueue#remove() is blocking:
http://download.oracle.com/javase/6/docs/api/java/lang/ref/ReferenceQueue.html#remove()

But I wonder: isn't this change making RequestFactory incompatible with
AppEngine?

There must be a better way.

IIUC, what WeakMapping is trying to do is a WeakHashMap where we
guarantee identity comparison (whichever the actual key, thus whichever
its equals() implementation). Maybe we should mimic WeakHashMap's
internal behavior then: each time you access it, poll() the
ReferenceQueue until it's empty.

http://gwt-code-reviews.appspot.com/1451819/

--
http://groups.google.com/group/Google-Web-Toolkit-Contributors

[gwt-contrib] Re: Ensure all ProxyAutoBeans not directly referenced from root object or via a shim can be garbage-... (issue1451819)

2011-06-16 Thread BobV 
 But I wonder: isn't this change making RequestFactory incompatible with
 AppEngine?

Reverting to original behavior of calling cleanup() from get()/set().

-- 
Bob Vawter
Google Web Toolkit Team

-- 
http://groups.google.com/group/Google-Web-Toolkit-Contributors

[gwt-contrib] Re: Ensure all ProxyAutoBeans not directly referenced from root object or via a shim can be garbage-... (issue1451819)

2011-06-15 Thread rjrjr 

LGTM


http://gwt-code-reviews.appspot.com/1451819/diff/1/user/src/com/google/gwt/core/client/impl/WeakMapping.java
File user/src/com/google/gwt/core/client/impl/WeakMapping.java (right):

http://gwt-code-reviews.appspot.com/1451819/diff/1/user/src/com/google/gwt/core/client/impl/WeakMapping.java#newcode104
user/src/com/google/gwt/core/client/impl/WeakMapping.java:104: for (;;)
{
why is this tight loop okay? What makes the thread sleep or whatever?

http://gwt-code-reviews.appspot.com/1451819/diff/1/user/src/com/google/web/bindery/autobean/vm/impl/ProxyAutoBean.java
File user/src/com/google/web/bindery/autobean/vm/impl/ProxyAutoBean.java
(right):

http://gwt-code-reviews.appspot.com/1451819/diff/1/user/src/com/google/web/bindery/autobean/vm/impl/ProxyAutoBean.java#newcode143
user/src/com/google/web/bindery/autobean/vm/impl/ProxyAutoBean.java:143:
* elsewhere. The alternative is a massive memory leak.
Don't understand this comment. How is it premature if nothing is
referencing them?

http://gwt-code-reviews.appspot.com/1451819/diff/1/user/src/com/google/web/bindery/requestfactory/server/RequestFactoryJarExtractor.java
File
user/src/com/google/web/bindery/requestfactory/server/RequestFactoryJarExtractor.java
(right):

http://gwt-code-reviews.appspot.com/1451819/diff/1/user/src/com/google/web/bindery/requestfactory/server/RequestFactoryJarExtractor.java#newcode67
user/src/com/google/web/bindery/requestfactory/server/RequestFactoryJarExtractor.java:67:
import
com.google.web.bindery.requestfactory.vm.testing.UrlRequestTransport;
Does this file belong in this patch?

http://gwt-code-reviews.appspot.com/1451819/

--
http://groups.google.com/group/Google-Web-Toolkit-Contributors