Refactor LoadGroupXMLFile inside LoadDetectorsGroupingFile

[Arturs Bekasovs]

In #7772 a parseRange function was added to Kernel::Strings, so parseRangeText is not needed any more. parseDetectorIDs() and parseSpectrumIDs() are really just parsing the output of parseRangeText, so these could be removed as well.

[Arturs Bekasovs]

Touches the same stuff. Better to update it all on one go.

[Arturs Bekasovs]

Return value optimization should help to avoid performance issues. Much more troubles working with shared pointers.

[Arturs Bekasovs]

We will not gain much anyway, so as soon as it works, leave it.

[Arturs Bekasovs]

Applies existing function and removed previously used ones.

Refs #7762

Changeset: 77f3137d9c8a4891d80715fe9411fba2b202b973

[Arturs Bekasovs]

Removed no longer relevant todo.

Refs #7762

Changeset: 3d50341cb146d493682f51839bdd05115077134c

[Arturs Bekasovs]

Tester:

Please check that test suite for the refactored class is rigorous enough and that it runs successfully. LoadDetectorsGroupingFileTest is what you should look at, as the LoadDetectorsGrouping uses LoadGroupXMLFile. The new function applied - String::parseRange - does have it's own test suite, you might want to look at it as well.

Check that the code changes seem reasonable.

[Karl Palmen]

Looks good. I find that the algorithm LoadDetectorsGroupingFile does what I expect.

To make the unit test more rigourous I suggest adding a function that tests that each group value gws->dataY(n)[0] is at least a great as the previous one gws->dataY(n-1)[0] and do a TS_FAIL if it finds the next one is of a lower number. Then with the extremes of the groups tested as at present one has then tested all the detectors.

If this slows down the test a lot, you can limit its use.

[Arturs Bekasovs]

Good suggestion. Shouldn't slow tests down a lot, as there is usually quite a limited amount of detectors/spectra in the test data.

[Arturs Bekasovs]

Improved unit tests. Now all the detectors/spectra are checked.

Refs #7762

Changeset: 90caaaf670487c11eed411e12b1ecccc9c65df9e

[Wenduo Zhou]

The improved unit test is good. Ticket is thus closed.

[Wenduo Zhou]

Merge remote-tracking branch 'origin/feature/7762_refactor_loadgroupxmlfile'

Full changeset: ​3fd9a31072217eeec1e4a8ab39be37a708a2b0a1

[Arturs Bekasovs]

(In #7716) Rethinking this, the pairing/naming etc. information is purely cosmetic stuff, which helps the users of MuonAnalysis to find the right groups and pairs quickly. We don't really need to store this information in a workspace, because none of the algorithms actually need it, they operate with group numbers / detector lists only.

That's why, it seems perfectly fine for the code to load/save that stuff to be a part of MuonAnalysis interface.

[Stuart Campbell]

This ticket has been transferred to github issue ​8607