On Linux, the gtk table seems to resize itself automatically when the
layout is computed and forces the last column to resize itself at +18 pixels.
This has a problematic impact on this dialog as it is not resizable and is
made to match the size of the content. As the dialog is used time after time
the last column grows by 18 pixels at every use up to a point where it
doesn't fit the screen.
Since storing the columns size is not that useful and I couldn't find a
way to ignore this first resize, I'm just removing the width storage.
ro.build.version.codename is a new property indicating whether a platform
is in its release form (value = REL) or in development (value = dev branch
name such as Donut). When the codename indicates a development/preview version
then the API level must be ignored and this codename is used as a unique
identifier of the platform.
IAndroidTarget has been changed to return an instance of a new class
AndroidVersion instead of the api level directly. This class helps deals with
the logic of comparing version from targets or devices.
This change impacts all of the sdk manager to deal with targets identified by
codename instead of api level. This in turn impacts everything that relies
on the sdkmanager: ADT (build, launch, project creation), the AVD manager,
the SDK updater.
- the "manager" button on the AVD selector (when not in manager
will open the Android SDK window
- the AVD manager in the eclipse toolsbar open the Android SDK
window as well.
Still left to do:
- remove the now obsolete AVD wizard
- figure out what to do with the other pages.
- New/Delete/Info/Refresh/Manager buttons.
- Ability to set any kind of AVD filter
- Self-refresh: Selector knows the AvdManager and can refresh itself.
- Manager mode or simple (selection or checkboxes)
- Manager mode displays broken AVDs.
Updated the SdkUpdate AVD page and the various use of the Selector in ADT.
Can't reproduce the NPE in ContentAssist listed in the
bug. However if things go really wrong there are a couple
objects that can be null so let's be defensive about them.
First, the ApkBuilder didn't cancel its run if the project had markers from JDT.
Second, the try/catch on ApkBuilder#build didn't properly test the
CoreException status severity (used getCode instead of getSeverity), so it did
not detect cancels being thrown by #abortOnBadSetup and displayed the error
in the console instead.
Basically:
editors.resources.manager -> resources.manager
editors.resources.configurations -> resources.configurations
This is to make it less confusing between the "Resources editors" and the
class parsing/handling Android resources (either in a project or in the
framework).
Also moved the ResourceExplorerView out of the resources editors, and clean
up a few other misc classes.
