Usability issues with GenerateEventsFilter

[Owen Arnold]

• Remove UseParallelProcessing as a property
• Remove NumberOfThreads as a property
• Documentation doesn't include property options in property descriptions or anywhere else. For example, FilterLogValueByChangingDirection doesn't let me know that Increasing and Decreasing are valid options without using the MantidPlotGui to open the algorithm dialog.

Other suggestions:

• Flexibility around StartTime and StopTime is actually confusing. For example if you know that StartTime is an offset in seconds, then want to be able to provide this as a number not a string.
• Roman thinks that the MatrixWorkspace output type is unnecessary and that the TableWorkspace could be used to do this very quickly.

[Martyn Gigg]

Is this a new algorithm? If not, we need to be careful when removing properties as it will affect user scripts.

[Owen Arnold]

No, this is not a new algorithm. The approach I've seen used (can't remember where) is to effectively mark the property as deprecated, so that anyone setting it to anything other than default will get a deprecated property warning. Then the properties should be removed in an iteration or so time.

[Martyn Gigg]

That would be fine, I just wanted to make sure it was flagged.

[Wenduo Zhou]

I think it better to keep property 'UseParallelProcessing' now. The complexity of the algorithm is O(N). In most cases, the length of the sample log is not huge. The sequential code will work efficiently. If parallelization is imposed on these, most of the computation time may be just used by the overhead of OpenMP. So if we drop the property 'UseParallelProcessing', then we will have to make the algorithm to judge whether parallelization is used.

[Wenduo Zhou]

This is about Roman's suggestion on MatrixWorkspace output. I did some simple tests and found that the computation time to use TableWorskpace and MatrixWorkspace are close. But I still prefer to keep the option for MatrixWorkspace output, because it is quicker to plot the target group against time from MatrixWorkspace than TableWorkspace. At least, I may need this feature some time.

[Wenduo Zhou]

Refs #10069. Corrected run end and log end.

1. Run end time is redefined in order not to miss the events in last

pulse;

2. Time series property is extended to run end;

On branch feature/10069_improve_geneventfilter

(use "git reset HEAD <file>..." to unstage)

modified: ../Mantid/Framework/Algorithms/inc/MantidAlgorithms/GenerateEventsFilter.h modified: ../Mantid/Framework/Algorithms/src/GenerateEventsFilter.cpp modified: ../Mantid/Framework/Algorithms/test/GenerateEventsFilterTest.h

Changeset: b7cfc8e33bfbcb27240c2a9d2582a545bfbc651b

[Wenduo Zhou]

Fixed a bug due to inaccurate run end time in log. Refs #10069.

Changeset: 68b32e92d97b621677276bb552c181a4411a3707

[Wenduo Zhou]

Refs #10069. New golden file.

Changeset: 326a869462cf0d643879d1d0e723990c2ff876c0

[Wenduo Zhou]

For tester

1. Issues described in #10022 is solved in this ticket too.

Run the script given in that ticket (​http://trac.mantidproject.org/mantid/ticket/10022) to see whether the new outcome is correct;

2. Check the unit tests and system tests;

3. Check the reply on the suggestions for improvement to see whether they are reasonable explanation.

[Wenduo Zhou]

Refs #10069. Improved some documention on properties.

Changeset: 38fe41da5cb1644ccd4bf8b31bc227eb47a9c8e0

[Vickie Lynch]

Merge remote branch 'origin/feature/10069_improve_geneventfilter'

Full changeset: ​33e0b7a819fa8b1ae6241db0579dc7d3f9371978

[Martyn Gigg]

This is being verified as ​pull request #5.

[Martyn Gigg]

Merge pull request #5 from mantidproject/feature/10069_improve_geneventfilter

New golden file for HYSPEC

Full changeset: ​0aeae2b42cd7b23d9e10478897ac1ae4fc416138

[Stuart Campbell]

This ticket has been transferred to github issue ​10911