Summary: | Eliminate duplication of JDT UI test helper classes | ||
---|---|---|---|
Product: | [Eclipse Project] JDT | Reporter: | Francis Upton IV <francisu> |
Component: | UI | Assignee: | JDT-UI-Inbox <jdt-ui-inbox> |
Status: | ASSIGNED --- | QA Contact: | |
Severity: | enhancement | ||
Priority: | P5 | CC: | bokowski, daniel_megert |
Version: | 3.5 | Keywords: | test |
Target Milestone: | --- | ||
Hardware: | All | ||
OS: | All | ||
Whiteboard: | |||
Bug Depends on: | 307423 | ||
Bug Blocks: |
Description
Francis Upton IV
2009-02-02 15:49:46 EST
There are pros and cons. You listed the pros (to which I agree). The cons for us are: - we get stuff we don't want (e.g. extra plug-in started ==> perf impact) and extension contributions that can impact our tests. Currently there are just two perspectives but we don't even want/need them and someone might put other stuff in there. - we have to compare all the helpers one line by one because the slightest difference in the implementation can have performance impacts (of course on both sides of the scale) on our performance tests. I would accept your suggestions under the following conditions: - this is not done by us (as we don't have time for this) and the person who does the work has to ensure that there is 0 performance impact i.e. we want numbers for every single test before we go ahead. This should be in the form of perf results that we can compare. - plugin.xml gets removed from org.eclipse.ui.tests.harness and will never come back to prevent future side-effects - org.eclipse.ui.tests.harness.UITestHarnessPlugin. It is currently not used and basically dead useless code which in the future can cause side-effects if people start to use it (e.g. adding stuff to start(...) - the bundle is turned into a non-singleton bundle |