Henri Sara [Mon, 7 Aug 2017 07:05:44 +0000 (10:05 +0300)]
Exclude Firefox from JavaScriptPreloadingTest (#9784)
The test works on Firefox under low load, but not when
the browser VM is slow due to load, in which case Firefox
sometimes executes scripts out of order. This seems to be
due to a Firefox bug - the generated HTML looks correct.
Fix RadioButtonGroup selection updates to client (#9749)
This patch provides a simple fix for the majority of issues. There are still issues that should be fixes by refactoring parts of the logic in AbstractSingleSelect.
This patch does not unify the handling of empty values in the TestBench elements of various AbstractSingleSelects.
Artur [Mon, 24 Jul 2017 05:43:52 +0000 (08:43 +0300)]
Do full connector tracker cleanup when the session lock is released (#9707)
As there is no "request end" call after invoking UI.access() from a background thread,
the connector map was not earlier properly cleaned afterwards. If you toggled visibility of a
component from the background thread, the tracker state became inconsistent.
If this becomes a performance problem, it could probably be optimized to that cleanup
is done in request end and only at the end of access if not inside a request.
Dos Moonen [Wed, 19 Jul 2017 06:24:07 +0000 (08:24 +0200)]
More user friendly fail message (was NullPointerException) (#9665)
Yes, I should never have created "D:\eclipse workspaces" in the first
place.
No, I am not replacing "%20" with spaces, or creating an URI instance.
"D:\eclipse%20workspaces" is a valid folder and I'm not willing to go
down the rabbit hole and see what that turns into.
Tim Ward [Wed, 19 Jul 2017 06:17:46 +0000 (07:17 +0100)]
Tidy up the Vaadin OSGi whiteboard component (#9648)
The Vaadin OSGi integration component uses Declarative Services, but it does some odd things:
* It uses the whiteboard service's own bundle context to get hold of the service instance
* It has an asymmetric get/release for the whiteboard service, which could leak instances over time
* It releases service instances that it is still actively using
This change tidies up the service lifecycle by delegating the get/release to the Service Component Runtime managing the container (specifically by injecting the service instance). Using this injection also ensures that the Vaadin whiteboard service is obtained using this component's bundle context.
This change also simplifies the code a little by using the reference as the key to track the registrations. Different references for the same service are required to be equal so there is no issue with doing this.
This change does not alter the fact that the whiteboard service's bundle context is used to register the Http Whiteboard servlet, as this may be the intended behaviour. As a result the component should be prepared for an IllegalStateException when unregistering the service whiteboard service, which may already have been unregistered by the OSGi framework if the target bundle is stopping.
Andrew Spear [Mon, 17 Jul 2017 11:03:07 +0000 (06:03 -0500)]
Change sendMessage access modifier to protected. (#9622)
This allows extending AtmospherePushConnection, and then doing overriding sendMessage. One use case is to correlate the IP address and the message in the logs.
Use ZoneId rather than ZoneOffset for conversion (#9620)
Use ZoneId rather than ZoneOffset for conversion between model (Date) and presentation of LocalDateTime.
LocalDateToDateConverter (correctly) uses ZoneId, whereas LocalDateTimeToDateConverter (incorrectly) used a ZoneOffset. This fix aligns the two Converter implementations and makes the latter one more robust.
A ZoneOffset is a fixed time difference, e.g., +07:00, whereas a time zone - represented by a ZoneId - is more dynamic, including features like Daylight-Savings Time. A ZoneId returns one or more ZoneOffsets via its ZoneRules method. (ZoneOffsets have trivial ZoneRules that simply return the ZoneOffset.)
Since the date/time being displayed may be from any date on the calendar, the ZoneOffset imposes a negative limitation. Using ZoneId instead gets us past that limitation and allows a more robust set of conversion rules.
Pekka Hyvönen [Thu, 29 Jun 2017 07:01:05 +0000 (10:01 +0300)]
Fix invalid index in drop target Grid (#9604)
Was using length - 1 and 0 for getting the row element,
which caused client side exception when dragging over header or footer when the grid had enough rows.
This had no effect on UX, only exceptions thrown. Now using visible range instead.
This patch changes Grid Columns, so they can have different value and presentation types. A presentation provider can be given when setting the renderer for a column. This provider takes the value of the column on a row and chooses what to present for this value. Using this approach it is easier to have an editor for the actual backing data instead of the presentation of it.
Leif Åstrand [Wed, 21 Jun 2017 08:02:46 +0000 (11:02 +0300)]
Load the debug window implementation asynchronously (#9026)
Reduces the gzipped DefaultWidgetset eager bundle size from 345kb to
325kb. The relative impact is greater optimized widgetsets that might
have down to around 150kb in the eager bundle.