ISIS SANS reductions are being affect by past reductions

[Steve Williams]

From Richard Heenan: If a value in the gui box for transmission spectrum is entered manually, it seems to be ignored! – another bug! Eg try entering 1, then the transmissions should come out at 1.0, they don’t, but give the normal values.

The solution needs to check that these things are always true:

• After running the SANS GUI once and changing settings: all settings that were changed must be reflected in the workspace history of the new output workspace
• Everywhere the method workspaceExists(workspace) is used in the Python code under Code/Mantid/Scripts/SANS the user settings that affect workspace must be checked too (e.g. from it's Run object)
• All execute() methods in the Reduction class must be "const" in that they don't change the class's data
• As the execute() methods can't be exception proof all exceptions within them must raise to a finally block that removes the reducer object and all workspaces touched by that instance of the reducer object
• All workspaces produced by a SANS run (e.g. trans_can_2.0_14.0) should but documented somewhere accessible from the links under http://www.mantidproject.org/SANS#ISIS . At least the use and the names of the algroithms used to create them must be included. Some detail on the settings used in their creation would be helpful

[Steve Williams]

(In [10503]) Add a method to reducer to remove all the workspaces it's create e.g. because of an error re #2757

[Steve Williams]

(In [10506]) Disable the caching of ISIS SANS transmission calculations because it opens the possibility of using the wrong files re #2757

[Steve Williams]

(In [10633]) Read the monitor from the SANS user file correctly re #2757

[Steve Williams]

The LoadRun steps, GetSampleGeom and GetOutputName execute() methods can't be const

[Steve Williams]

(In [11023]) SANS ReductionStep.execute() methods must be "const", except the LoadRun steps, GetSampleGeom and GetOutputName execute() methods can't be re #2757

[Steve Williams]

(In [11040]) Rapid simple fix of broken system test re #2757

[Martyn Gigg]

(In [11042]) Avoid a bug in deepcopy with Python 2.6 which stopped me running the system tests in Ubuntu. Refs #2757

[Nick Draper]

Accepted and assigned tickets moved at iteration 29 code freeze

[Steve Williams]

In [13119]:

The cutoffs passed to Q1D weren't getting reset to zero when is WCUT and RCUT aren't in the mask file re #2757

[Steve Williams]

In [13573]:

Removing some potentially dangerous code re #2757

[Steve Williams]

• The transmission and other settings mentioned in the first point seem to be going through.

• On point 2 workspaceExists() is only being used as a check before DeleteWorkspace, which is OK as it doesn't allow old workspaces to servive into a new reduction

• Point 3 doesn't now need to be inforced as the reduction is copied to the varaible current_settings (in isis_reducer) before execution could change things

• Point 4 The highest level is in ISISCommandInterface.py and this has try: finally pairs around functions that call _reduce(). However, deleting partially complete workspaces isn't implemented, we'll wait for a request from the users.

• There is documentation for the workspaces in mantidproject.org/SANS_Flow_Diagram at the bottom of the http://www.mantidproject.org/SANS#ISIS section

[Stuart Campbell]

This ticket has been transferred to github issue ​3604