| Summary: | [TableViewer] Deactivation of CellEditor resets selection | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [RT] RAP | Reporter: | Steffen Eichenberg <steffen.eichenberg> | ||||
| Component: | JFace | Assignee: | Project Inbox <rap-inbox> | ||||
| Status: | RESOLVED FIXED | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | P3 | CC: | ivan, ruediger.herrmann | ||||
| Version: | 1.4 | ||||||
| Target Milestone: | 1.5 M5 | ||||||
| Hardware: | PC | ||||||
| OS: | Windows 7 | ||||||
| Whiteboard: | |||||||
| Bug Depends on: | 366738 | ||||||
| Bug Blocks: | |||||||
| Attachments: |
|
||||||
|
Description
Steffen Eichenberg
Tested again without 'virtual'. Problem exists only with 'virtual' tables. I cannot exactly reproduce the problem. After deactivating the cell editor and pressing arrow-down, the 5th row is selected as expected as can be seen in the "Selection:" label on the right. What doesn't work reliably though is the selection highlighting. Most times, the active cell is missing the bule background color. Can you confirm that or am I missing anything? Opened bug 366575 for what is desribed in comment #2 I can reproduce the described bug with RAP 1.4.0 and RAP 1.5.0M3. Today i downloaded a fresh copy of RAP 150M3, setup a new target platform in Eclipse, extracted the controls demo, started it with JDK 1.5 (Sun, Windows 7, 32 Bit), opened the controls Demo in FF 8.0 and made a short screencast of the effect. You can find the screencast under the following address (valid for 7 days): http://www.screencast-o-matic.com/watch/cXl2iUvGA The bug manifests itself only with 'virtual' tables. Regular tables do not have that problem. I can reproduce what you describe with 1.5M3 but not with CVS HEAD. Could you give a latest nightly build a try (http://www.eclipse.org/rap/downloads/)? With migration of Table to the JSON protocol we changed (seems for good) the selection rendering. Probably, this fixes the issue. Created attachment 208417 [details]
Proposed fix
Changes virtualSetSelectionToWidget() so that it calls doSetSelection() when the 'reveal' parameter was set to true, otherwise it calls doDeselectAll() and doSelect()
For brevity of this patch, the now obsolete tracking of the 'firstItem' was not removed. This should be done in a final version.
@Ivan: I have tested it with the NB Version of today (15.12.2011, rap-runtime-1.5.0-N-20111215-0213.zip). Problem still exists. (In reply to comment #7) > @Ivan: > > I have tested it with the NB Version of today (15.12.2011, > rap-runtime-1.5.0-N-20111215-0213.zip). Problem still exists. Yes... that's true. Yesterday we found that it worked with HEAD because of a not so correct code. I removed this code and now HEAD works the same with M3 and 1.4.1. The correct fix is outlined in the patch from comment #6. Applied patch from attachment 208417 [details] to Tree_Table_Merge branch
Applied patch from attachment 208417 [details] to HEAD as an internediate solution until bug 366738 is resolved. |