TRACKING: post-construction 2/7/24

[cmarooney-stfc]

Agenda modification for this meeting:

It was agreed at the last sprint review meeting that this meeting should discuss planning for the period after closeout. As David cannot be present, this discussion will in fact take place at a separate meeting next week, to ensure everyone can be present.

Reports and actions:

Jacob:

• Horace/SpinW interface

JW has discussed with Duc the long-term support for the Horace/SpinW interface, and it was agreed that Richard and Duc will take over these interfaces for maintenance and further development.
A protocol has to be decided on how this will work - the Horace and SpinW teams will discuss this, with a view to deciding how this should be achieved, including testing.
JW will be the Horace liaison for this.
It was noted that we have existing CI to track such development.

• Graduate project proposal

The proposal for this is ready, and has been approved by Anders.
ISIS can fund this if SCD is not able to. As SCD has only 8 ALC spots to provided all its graduate funding, we propose that ISIS should plan to fund this immediately without attempting to apply to SCD.
It was noted that this is closely related to Alex's current student project. Alex and Jacob should work on preparatory work, to narrow the possible options, and de-obscurify.
A submission should be planned to be made by 22/7, for an Aug/Sept start.

• Machine configuration manager

A prototype configuration manager is now ready. It is based on actual detected machine features, rather than having predetermined list available to choose from. It is now being documented, with tests being added.

• Actions for JW:

  • Finish configuration manager, including any testing and debugging. After this it should be passed to Ross for user testing.
  • • Finish HPC parallel settings for Horace-4 as part of the machine configuration work. It was noted that Horace-3 can be left unchanged.
  • Start scoping the graduate project, which should specify the project in detail for the graduate. This should be complete by the time Alex returns from leave. The proposal should then be submitted.
  • Fix the prototype for the persistent cluster.
  • Meet with Duc and Richard for the work protocol for the Horace/SpinW interface.
  • The issue of the Horace/Euphonic interface CI failures should be progressed. In this context, it was noted that single-sign-on access to the Matlab file store have meant that the files stored there under euphonic@stfc.ac.uk are not longer simply accessible as we cannot enter the password we have. JW has plans to try to "hack" this but otherwise Matlab and/or UKRI should be approached to get round this.

Alex

• Validated grid intersection algorithm

This has been validated and is fine.

• cut-dnd

This has been disable due to concerns about its behaviour. It was noted that Ross expects normalised polygons behaviour from Mantid. The existing axis-aligned cut-dnd works, jhowever the projection version doesn't, due to bin subdivision. We agreed to park the cut-dnd issue until after closeout, when a use-case will have been formulated.

• Actions for AB:

  • Finish replicate in parallel.
  • Clarify what the issue for run-id is, so we can progress it.
  • The student project shoul, given the amount of time available and the abilities of the student, confine itself to a feasible amount of refactoring and testing of combine_sqw.

Chris:

• Instrument info from nxspe files:

The existing code for constructing instruments from nxspe information is largely working. Deficiencies in the information now in the nxspe files are: (a) moderator info overwritten (b) additional rep_* records need to be ignored as the format causes errors. Passed to Duc for fixing.

• Actions for CM:

  • Ensure David and Ross arrange a post-mortem meeting on Horace-4 in the style of the previous meeting to provide a user view on the current state of Horace-4.
  • Obtain a ustomer report from David for the closeout report
  • ??? cut-dnd question

Toby;

Toby's current task is to create processing for the detector data in the nxspe files to reduce it to what is needed to populate IX_detector_arrays. This is in progress; as Toby had to leave the meeting early there is no additional report on this.

---

Here are the original notes from the meeting; above is the edited and (hopefully) polished version (CM):

David not coming this time - discussion of post-closeout needs will be next week

Jacob:

Richard and Duc will take over interfaces for SpinW/Horace for maintenance and tracking of evolution
A protocol has to be decided on how this will work - PACE and SPinw to discuss (=decide, test and approve)
Jacob to be soPACE liaison
CI exists

Grad project proposal ready, approved by Anders
ISIS to fund if not SCD
SCD has 8 ALC spots - propose ISIS funding
related to Alex student work
Alex/Jacob should work on preparatory work - narrow options, de-obscurify
submission by 22/7 - Aug/Sept start

prototype config manager ready based on machine features rather than predetermined list- now being documented with tests being added

Actions:
finish config manager (test/debug/Ross test)
start scoping for grad project - specify details for grad (by end Alex leave)
submit proposal
fix persistent cluster prototype
meet with Duc and Richard
finish HPC parallel settings for H4 as part of configuration, H3 can be left (same as above)
deal with Euphonic CI issue
Euphonic interface issue - single-sign-on changes access - need to figure out pw for euphonic@stfc.ac.uk

Alex

Validated grid intersection algorithm - is fine
cut-dnd - disable due to concerns - Ross expects normalised polygons behaviour from Mantid
existing axis-aligned cut-dnd works, projection version doesn't, due to bin subdivision
park cut-dnd issue until post-closeout when a use-case has been formulated

Actions:

replicate in parallel
run-id - need to clarify what this actually is
student project - refactor and test combine_sqw

Chris

Actions

Harass David/Ross for H4 post-mortem - they do meeting and report to us
Customer report from David
cut-dnd question