Long term handling of embedding Nexus IDFs

[Anders Markvardsen]

As approved by TSC implement suggestion 2 in ​https://github.com/mantidproject/documents/blob/master/Design/EmbeddedInstrumentInfoNexus.md

[Karl Palmen]

After a talk with Anders, it was decided to go ahead with the second version of Suggestion 2.

What we currently do can be expressed as

Load Embedded IDF
If(Embedded PF) load it
else if (external PF) load it

This would be changed to

# Some of Existing part
Load Embedded IDF

# New part
if( Correction file) 
   load it
   fn = ""
   ap = true
   for (each correction_item) 
      if(rawfile.date in correction_item.dates) 
         fn = correction_item.filename
         ap = correction_item.append
   if(fn)
       if(ap == true)
          # from existing part
          If(Embedded PF) load it
          else if (external PF) load it 
        load fn
   else
       # from existing part
       If(Embedded PF) load it
       else if (external PF) load it 
else
   # from existing part
   If(Embedded PF) load it
   else if (external PF) load it 

This assumes that the date ranges of the correction items do not overlap and does not explicitly attempt to detect such an overlap. Instead, this can be done when the correction file is loaded and an exception thrown if an overlap is found.

[Karl Palmen]

The implementation of

If(Embedded PF) load it
else if (external PF) load it 

appears to be split between two or more algorithms. The details of this implementation need to be found out before work can proceed on this ticket. It appears that only SNS make use of the second line and ISIS use only embedded parameters if the IDF is embedded.

[Karl Palmen]

I've found out that

Load Embedded IDF
If(Embedded PF) load it
else if (external PF) load it

is implemented in LoadIDFFromNexus, which is called by LoadISISNexus2. Hopefully, this is the only place it is implemented.

[Karl Palmen]

Refactoring is necessary. Both

Load Embedded IDF
If(Embedded PF) load it

are done by ExperimentInfo::LoadInstrumentInfoNexus. These two need separating to allow the loading of the parameters to be deferred and possibly replaced by a different parameter load.

[Karl Palmen]

ExperimentInfoTest does not have a unit test for ExperminentInfo.loadInstrumentInfoNexus .

[Karl Palmen]

Planned work on this ticket

(1) Unit test ExperminentInfo.loadInstrumentInfoNexus (#11719).

(2) Make separate functions to Load embedded IDF and Load embedded parameters (currently both done together by ExperminentInfo.loadInstrumentInfoNexus).

(3) Make one function to Load embedded parameters or if not found load parameter file. This is currently done separately in ExperminentInfo.loadInstrumentInfoNexus and LoadIDFFromNexus.

(4) Function or code to search for and find the XML file that has the corrections to parameters in it, henceforth referred to as the correction file. This must not take a long time if no such file exists.

(5) Function or Algorithm to read the correction file and output a filename & append flag as required.

(6) Code that makes use of the products of the previous steps to implement the ticket in each loader that may make use on an embedded IDF (may be one step per loader).

[Stuart Campbell]

This ticket has been transferred to github issue ​12255