Release Notes 3.2.1

Patch Release
Version	3.2.1
Requested by	Nick Draper
Approved by	Peter Peterson
Authorised on	14-Aug-2014
Code Reviewed by	Martyn Gigg
Code Reviewed on	18-22 Aug 2014
Review Recommendation	Release
Released on	22-Aug-2014

Key Features

This is a patch release to Mantid that corrects some significant issues since the last full release.

There are instrument definition updates for CNCS & TOPAZ and a critical bugfix to LoadISISNexus to correct a problem with assigning spectrum numbers to the correct detectors when loading a partial file. The ISISLiveEventListener has also been improved to support a greater event rate. In addition there have been a few more minor bug fixes.

Citing this release

Mantid (2014): Manipulation and Analysis Toolkit for Instrument Data.; Mantid Project.
http://dx.doi.org/10.5286/SOFTWARE/MANTID3.2.1
Full author list

Changes in this version

Ticket	Summary	Impact	Solution
10012	__file__ not propagated in Script Window	Scripts should be able to rely on this variable being present so that they can find their own location	Copy variable between dictionaries before clearing other variables on 'clean' runs
9956	SURF Parameter Fix	Refl GUI gives error in stitched data due to incorrect parameter value in the IDF parameter file	Fix the parameter value
10095	Improve ISIS live data listener performance	Ensures that the ISIS live event listener can keep up with event rates in white-beam mode on big instruments	Decreases sleep times and cuts down on number of socket.sendBytes() calls.
9671	Create a TEST_LIVE facility	Allows users to test out the live data algorithms without using a debug version of Mantid	Adds a TEST_FACILTY to the facilties XML file
9798	SliceViewer Exception For Infinite Dimensions	Avoids an unhandled std::bad_alloc exception when using the SliceViewer	Check the dimension limits before attempting to display the workspace
9971	Hang/crash with debug logging in SofQW3	Allows SofQW3 to be run with at debugging log level	Raises logging statement out of a tight loop in favour of accumulating the messages
9887	Crash with Slicing Algorithm Dialogs	SliceMD/BinMD dialogs are now stable again for users	Remove a redundant signal-slot connection
9539	CCL Clustering Defect	Some pixels are misplaced when deciding what is connected	Fix the IDs of the element labels
10113	Error in ISIS REFL Gui with No Angle Specified	ISIS Refl GUi can be used by entering only a run number	Explicitly check for empty angle cell
10088	Incorrect detector IDs in LoadISISNexus when loading limited spectra	When selecting a subset of spectra to load, LoadISISNexus does not assign the correct detectors to the loaded spectra.	Update spectrum/detector number assignments using the map rather than just counting from 1
9980	New WISH IDF	Allow extra panels to be seen in Mantid	Add new IDF file and update dates of older files
9961	4D MD Workspace Always Converted to Same Binnng	MD binning can now be set at the conversion time so that in VSI you don't have to rebin as soon as you start	SplitInto parameter is now carried through the MD conversion in ConvertToMD
10097	Bug in ISIS Refl GUI Resolution Calculation	ISIS Refl GUI now runs successfully when calculating the resolution	Theta parameter is always converted to a float
9954	Speed up Stitch1D	Stitch1D should be notably faster	Some internal function calls were profiles and sped up
9955	Nans in Stitch1D	Nans should not propagate through Stitch1D	The nans/infs are replaced by zeroes and then put back after the processing has been done
10104	ConjoinWorkspaces does not fix spectrum number for CheckOverlap=False	Using ConjoinFiles on several different runs produces different spectrum numbers	Fix up spectrum numbers when CheckOverlap is not requested
9335	GenerateEventsFilters and FastLog=True crashes for integer logs	GenerateEventsFilter can now be used on either double/integer logs using the FastLog option	Further checks are added for log types within the code and the appropriate action taken.
10028	GenerateEventsFilters does not filter correctly for separate directions	The sum of events from separate Increasing/Decreasing log values is now the same as those filtered by a filter created with the 'Both' option	Fix problems with identifying the direction of change and boundary conditions
10132	New CNCS IDF	Allows CNCS to run the new cycle with Mantid	New IDF file and the old has been renamed
10138	New TOPAZ IDFs	Makes different detector banks visible	New IDF file and the old ones have been renamed
10154	Fix CNCS IDF	Fixes a bug in CNCS geometry	Corrects the X/Z positions of some banks

Code Review

Review of Specific Changes

Ticket	Summary	Implementation Notes	Probability of side effects	Merged
10012	__file__ not propagated in Script Window	In 3.2, when executing a whole script, all of the variables definitions are cleaned out to ensure a that previous runs do not affect the code. The clear out was also removing the already defined __file__ variable. The solution simply makes a special case for it and copies it from the old scope to new.	Medium	18 Aug
9956	SURF Parameter Fix	The value in the Parameter XML file has been updated to that requested by the instrument scientist. This only affects the ISIS REFL and did not break any system tests.	Low	18 Aug
10095	Improve ISIS live data listener performance	Decreases sleep times and cuts down on number of socket.sendBytes() calls by accumulating events in an array and sending a batch in a singleshot. Also adds a basic test facility.	Medium	19 Aug
9671	Create a TEST_LIVE facility	A TEST_FACILTY has been created with various instruments to support testing the live data algorithms. There are a fair number of changes across a range of files. They are mostly fake generator specific and the changes but there are minor changes to the ISIS histo & event listeners to use a different port and these have been heavily code reviewed. Note that this dependent on 10095 as the basic TEST facility was created there first.	Medium	19 Aug
9798	SliceViewer Exception For Infinite Dimensions	The limits for the dimensions attached to the sliceviewer are now checked for inf and nan values	Low	19 Aug
9971	Hang/crash with debug logging in SofQW3	Debug logging was done for each bin in a tight loop but it has now been moved out of the loop so that the log messages are accumulated in a stream and then sent to the logger once per spectrum	Low	19 Aug
9887	Crash with Slicing Algorithm Dialogs	In 3.2 the dialogs became non-modal and calling accept() now executes the algorithm. The SlicingDialog had a double call to accept and so tried to execute the same algorithm object twice, leading to undefined results. The fix was simply to remove the redundant connection.	Low	19 Aug
9539	CCL Clustering Defect	DisjointElement nodes now have the ability to reset their ID and copy constructing one does not update the parent, as it was previously. This ensures that each neighbour has the correct ID & parent.	Low	19 Aug
10113	Error in ISIS REFL Gui with No Angle Specified	Explicitly check for empty angle cell in isis_refl_gui.py and pass None of the cell is empty	Low	19 Aug
10088	Incorrect detector IDs in LoadISISNexus when loading limited spectra	The fix is the minimal required to keep existing behaviour but also cover the additional loading of selected spectra. Tests have been put in place that will now continue to verify that this is correct.	Medium	20 Aug
9980	New WISH IDF	Adds new IDF file with the dates set corrrectly and updates valid from/to dates for other definition files	Low	20 Aug
9961	4D MD Workspace Always Converted to Same Binnng	Fixes are minimal in CovertToMD and MDWSDescription. Additional unit tests have been added to cover these cases. There are also changes to systemtests because now the reference results have the correct binning rather than the default 10x10x10x10 but I see this as acceptably low risk.	Low	20 Aug
10097	Bug in ISIS Refl GUI Resolution Calculation	Theta parameter can sometimes be a string so it is now always converted to a float before use in calculations	Low	20 Aug
9954	Speed up Stitch1D	The code was profiled and found that some internal function calls were slow and created a lot of waste. These have been improved. This was not a requirement for the patch but #9955 is and it is dependent on this.	Low	20 Aug
9955	Nans in Stitch1D	The nans/infs are replaced by zeroes and then put back after the processing has been done	Low	20 Aug
10104	ConjoinWorkspaces does not fix spectrum number for CheckOverlap=False	New behaviour is that the spectrum numbers are fixed up if CheckOverlap is turned off. This makes sense as things with the same spectrum number aren't going to be that usable.	Medium	21 Aug
9335	GenerateEventsFilters and FastLog=True crashes for integer logs	Further checks are added for log types within the code and old statements that did nothing for integer logs have been implemented.	Medium	21 Aug
10028	GenerateEventsFilters does not filter correctly for separate directions	Fix problems with identifying the direction of change and boundary conditions. There are a large amount of code changes but the refactoring has helped with readability and the tests have been improved to cover these situations.	Medium	21 Aug
10132	New CNCS IDF	Generated a new IDF from python mantidgeometry scripts	Low	21 Aug
10138	New TOPAZ IDFs	New IDF file and the old ones have been renamed with correct valid-from/valid-to dates	Low	21 Aug
10154	Fix CNCS IDF	Fixes a bug in CNCS geometry	Low	21 Aug

Are any changes included in the release that are not accounted for above?

No

Is the collective effect of the changes acceptably low risk?

Yes

Code Reviewer Recommendation

Release once kits have undergone some unscripted testing - Complete.