SANS Redesign Load Data

[Gesner Passos]

It is necessary to introduce a clear separation between what is used for processing the reduction (ReductionSteps) and data loaded. Currently, these entities are mixed up, and this introduces problems not only in the design, but also to manipulate these entities.

By breaking these entities we will be able to introduce the changes necessary to process event data multiperiod data asked by larmor.

[Gesner Passos]

re #8435: Separate Can and CanSubtraction

now, reducer has a _can_run that works similarly to the _sample_run, and CanSubtraction does only reduce the Can if it is present.

Changeset: 8508f1fd13b939e44b86a0163079c057b9376643

[Gesner Passos]

re #8435: Adjust usage of background_subtracter inside Mantid

Where it was being used, it is now corrected.

Changeset: 2c97efdb0045dc76847894a93d4e4ab711770e8c

[Gesner Passos]

re #8435: TransmissionCalc does not use anything from BaseTransmission

Remove the non-sense inheritance.

Changeset: 1738e5ea133d941a4ec6f28eaf49a1ba9ef341c4

[Gesner Passos]

re #8435: Separate LoadTransmission and TransmissionCalc

One is related to loading data and the other to is a reduction step.

Changeset: c72f2fdd0136a136ecd375bf01a49d7cbb46311f

[Gesner Passos]

re #8435: Express that loading data is not reduction step

Changeset: b9afd2e7a3a19ee5ff30ae8d39a3b8e041686436

[Gesner Passos]

re #8435: Impose that AssignSample will be always the first

This commit ensures that the reduction will never use can or transmission from previous time. It will help to the goal of removing the copy and creation of more than one singleton inside the current reduction.

Changeset: 76e8cf244b0bf15cded5d3553aec7569677aeafb

[Gesner Passos]

re #8435: SANS Reduction

this is a temporary comment to explain why this code still works. It is my intention to get rid of the _setUpPeriod method by applying a better way of ensuring the execution of multiperiod data.

Changeset: 6beb096c9b2e1ca0d92f2022329c99c776f25c64

[Gesner Passos]

re #8435: Fixing UnitTest SansIsisGuiSettings

Changeset: 4ead550ab3a4d8a59d1aa0594660e3216eb1608c

[Gesner Passos]

re #8435: Ensure geometry correction gets the correct information

Changeset: eb0beb186d1436938cdce23440c2b9e7ff19872e

[Gesner Passos]

re #8435: SANSReduction requires passing first AssignSample

The GUI does this, so, it does not break the GUI rule.

Changeset: ​https://github.com/mantidproject/systemtests/commit/652a4d69474bf237716bcd313c059973769026d7

[Gesner Passos]

Tester:

Before explaining how to test, I want to explain the reasons and what we will gain by having these changes.

• By providing a separation of that is data and what is reduction steps we will be able to pursue the objective placed in #8443 and #8444.
• It is also a question of design improvement. Do not mix-up entities.
• The steps now ask the reducer in execution time for the inputs, and this will allow to reuse the steps without recreating the singleton.
• There is two changes I'm not really happy with:
  • Why we can not execute the geometry steps for the CAN and we can not correct the geometry of the CAN using its own geometry. (comment:12). I've asked the scientits if this is correct. """The scientist confirmed it is correct."""

• I was forced to update the systemtest as well comment:13, because it was assigning Sample after transmission, which is not allowed anymore. This in fact will ensure that all the reductions will not be influenced by data of previous reduction.

[Gesner Passos]

Remember of update code and systemtest.

• Open ISIS
• load systemtests/Data/SANS2D/MaskSANS2DReductionGUI.txt
• change to batch mode
• load sans2d_reduction_gui_batch.csv
• perform the reduction
• change to single mode
• perform the reduction
• check that they are the same as systemtests/SystemTests/AnalysisTests/ReferenceResults/SANSReductionGUI.nxs
• If you type CheckWorkspacesMatch it may complain that the sample name is not the same, because one is and the other ' '. I've already raised a ticket for this.

[Gesner Passos]

I could not forgive myself to break an agreement and ship a bad code (comment:12). - I broke the agreement that variables starting by underscore is meant to be protected or private. Today I thought a new way of implementing it, and releasing my conscience. Besides, it does improve a little bit the performance by not calculating the same value twice.

[Gesner Passos]

The existence of many singletons does not allow me to solve the issue presented in comment:16 as I planed #8455. I will have to wait to get the #8443 sorted before.

[Gesner Passos]

re #8435: SANSReduction requires passing first AssignSample

The GUI does this, so, it does not break the GUI rule.

Changeset: 38991892e6366623291ec2c31d2d50c1b3a095d5

[Gesner Passos]

It is probably better to let Anders or Peter Parker to test this ticket.

[Peter Parker]

Merge remote-tracking branch 'origin/feature/8435_redesign_load_data'

Full changeset: ​9f24f65f00b0603c426fbeca35b8bb7c10154ffa

[Peter Parker]

Merge remote-tracking branch 'origin/feature/8435_redesign_load_data'

Full changeset: ​a53daa3c28d91a003c366ff5a52d52e4063b88f1

[Gesner Passos]

Merge branch 'feature/8435_redesign_load_data' into feature/8443_many_singletons

Full changeset: ​519339a03b2f131c871fe3805267acde57871d25

[Stuart Campbell]

This ticket has been transferred to github issue ​9279