Skip to content

Bypass: Rework pad alignment

Timothy Davies requested to merge tdavies__rework_pad_alignment into ldavis/bypass

Prereq !118 (merged)

Slight rework of initial pad alignment & fake history pushing. I haven't reinvented the wheel, just trying to take the reworks in bypass and combine them with postcoh in a way that both are working as simply as possible. I've listed all the changes in the commit message.

Where Bypass stores data per pad (with different offsets/lengths on each) and postcoh uses the same values for all pads, I've just copied that data to each pad in postcoh (somewhat redunantly). I figure at worst it keeps our options open to store different history lengths per pad in postcoh if ever we want that.

Postcoh also did additional work with 'fillin_discont' that bypass didn't do. That just takes any unexpected gap in the input streams (not data marked as gaps, but actual missing timespans where buffer offsets didn't line up) and fills it with 0s. Seems like a reasonable enough thing to do in bypass too, but one to keep an eye on. Bypass uses slightly different timing management to postcoh, but as far as I can tell it shouldn't matter...

Tests

This is definitely getting into the riskier form of refactor/rework. I've set a run going without bypass here:

/fred/oz016/tdavies/projects/testing/run_tdavies__rework_pad_alignnment/

But it should definitely do thorough tests with bypass.

Merge request reports