Rename Workspace Does Not Check Duplicate Name in Other Workspaces

[Karl Palmen]

I find that if I open several workspaces. I then rename one of them so its name is the same as one of the others. That workspace or the one being ranamed vanishes and no error message is given.

Ideally RenameWorkspace and RenameWorkspaces should detect such a situation and refuse to rename and issue an informative error message.

[Karl Palmen]

Another issue is that RenameWorkspaces cannot be used to swap the names of two workspaces. This would be complicated to implement and the user can do the name swap in two rename operations.

[Nick Draper]

You can do it in two rename operations and that is enough. The logic of swapping if the resulting workspace exists would not make sense to most people, and the need for this is rare at best.

[Arturs Bekasovs]

Added a new flag defining whether a workspace name should be unique within ADS.

Refs #7309

Changeset: 9a489740ad03134f8d18dfbd331edd84174b62bf

[Arturs Bekasovs]

Made error message more informative

Refs #7309

Changeset: d48b218c933a7456601fae5650f688c694fee9ee

[Arturs Bekasovs]

Updated rename algorithms so they don't allow using already existing workspace names.

Refs #7309

Changeset: ddd795383b973abfb35692e207c5e2ddb8c585e8

[Arturs Bekasovs]

Added the name of the workspace to the error message

Refs #7309

Changeset: 9047d32c9482efb859f446c09af483a82779b8a4

[Arturs Bekasovs]

Made the algorithm handle group workspaces correctly.

Refs #7309

Changeset: 39fa9c4882e5fa0159480928152788362a5dc7ba

[Arturs Bekasovs]

Updated the tests and added a few more.

Refs #7309

Changeset: 7267cb29449141f5d42e9fd8df772d85aec203f1

[Arturs Bekasovs]

Added explanation of what UniqueMode is

Refs #7309

Changeset: 4484dbd341f6dc6ffbd99a6bdd37d7e4b028b44e

[Arturs Bekasovs]

Added some tests for WorkspaceProperty

Refs #7309

Changeset: 98a17aec24dfc6416b3a1cb517a4c207d56460fb

[Arturs Bekasovs]

Fixed the problem with renaming in ADS.

This was causing tests to fail.

Refs #7309

Changeset: eac1866423e8a7f92bd4b48e50688a84726808ac

[Arturs Bekasovs]

Update documentation for the algorithms.

Refs #7309

Changeset: f07bacfa994f137b875c9c54d92169b7efd1f160

[Arturs Bekasovs]

Branch: 7309_rename_ws_duplicate_name

Tester:

Open any two workspaces.

Rename the first one to ws1, and the second one to ws2. Rename should be successful.

Then try to rename ws1 to ws2. Rename dialog shouldn't allow that, showing a red error star near the OutputWorkspace parameter. Point a mouse the star. An error "The workspace with name "ws2" already exists" should appear. Notice that both ws1 and ws2 are still there.

Then, open another workspace and rename it to new_ws1. Run the RenameWorkspaces algorithm, and set InputWorkspaces to "ws1, ws2" and a Prefix parameter to "new_". Then execute an algorithm. It should throw an error saying "The workspace with name "new_ws1" already exists". ws1 and ws2 should both remain unchanged.

Open RenameWorkspaces again, and try "other_" as Prefix this time. This should throw no error, ws1 and ws2 shoud disappear, and other_ws1 and other_ws2 should appear.

[Jay Rainey]

Testing

• Testing on Ubuntu 12.4.
• Successfully prevents the user from renaming a workspace to one that already exists of that name.
• No longer overwrites workspace if name is already existing.
• Good error messages are thrown when a workspace name exists that the user wants to use.

Code review

• Clean and consistent code.
• No visible bugs/problems.

[Jay Rainey]

Merge remote-tracking branch 'origin/feature/7309_rename_ws_duplicate_name'

[Arturs Bekasovs]

This ticket was done and tested, but the way it was done conflicted with how the Mantid fundamentally works and broke almost all of the system tests. All of the algorithms do rewrite their results no questions asked, and rename algorithms should not make an exception. All the commits were taken out of the master.

However, the problem still remains - when renaming the workspace it would be nice to make sure user realizes that he might lose some important data, so the confirmation might be a good idea.

Need to come up with a way of doing this and think about whether it might be useful for when user is running other algorithms as well.

[Arturs Bekasovs]

This ticket has quite a lot of [bad] heritage, so I've created a new one for solving described problem: #9161. Consequently, this one becomes invalid.

[Jay Rainey]

The functionality described in comment:21, and in #9161 will be fixed in #2480. Closing as invalid.

[Stuart Campbell]

This ticket has been transferred to github issue ​8155