This work appears to be blocked by DM-35718 .

Blockage by DM-35718 (PIPE2D-1102) no longer the case. However, as bias, dark and flat corrections need to be applied to b+r processing, but not for n, for the science processing two separate runs need to be made with reduceExposure.py: one for b+r using bias, dark and flat corrections, and one for `n` without bias, dark and flat corrections. This specific feature is currently not supported in the YAML command generation mechanism. sogo.mineo has been informed and looking into a means of supporting this.

We should simply disable the loading of bias and dark for n in our custom IsrTask. I think we want to keep the flats, though.

Disabling the bias and dark for n in IsrTask would be good. But just want to re-iterate (mainly for Mineo-san's information) that if the flats need to be configured separately for n, an update to the command generation mechanism is still needed.

I began working on extending the command generation mechanism.

Where is the I/O done? They are ignored in the processing.

Can I add a commit to the current tickets/PIPE2D-1088 branch worked on by Hassan? Is it legitimate act?

This is the commit I want to push:

generateCommands.py: Introduce subgroups of ScienceStep

After this patch is applied, the following construction (`group:...`)
will be valid.

scienceBlock:
  -
    name: process_brn
    id: ["visit=47", "arm=b^r^n"]
    policy:
      reduceExposure:
        config: ["isr.doDark=False"]
        group:
          -
            id: ["arm=b^r"]
          -
            id: ["arm=n"]
            config: ["isr.doBias=False"]

In this case, "arm=b^r" will be processed with `--config isr.doDark=False`
and "arm=n" will be processed with `--config isr.doDark=False isr.doBias=False`.

Sure sogo.mineo feel free to add commits to tickets/PIPE2D-1088 as you wish.

We shouldn't need to change anything in the command generation. We just need to tell the ISR that we don't need bias and dark for n. The flats will be required for n, just as for b and r.

I've changed the PfsIsrTask so it won't attempt to load the bias or dark for arm=n. This should allow us to process CCD and H4RG data with the same configuration.

OK. I'll incorporate your changes Paul in my changes to the weekly build, and get back to you if there's any issues.

This may no longer be necessary, but here is the branch I made yesterday (Hassan told me about personal branches): u/SogoMineo/20221109a

Converted n-band PFFA files (generated by 2D simulator) to PFFB format. ISR-corrected images were first generated, and then converted to PFFB format using script dev_2ddrp/scripts/convertPFSAtoPFSB.py. For details of conversion process, see notes in https://github.com/Subaru-PFS/dev_2ddrp/blob/366f2bad38321de16cafb3197163a2bf30f7f013/notes/pipe2d-1088-notes.md .

Code changes look reasonable; I've not looked at the data.

Merged to master