search

ijiraq / gemini_processing

scripts and stuff to process gemini data witin the CADC.
1 stars 1 forks source link

Create tests of the NIFTY container based on representative sets. #11

Closed ijiraq closed 3 years ago

ijiraq commented 4 years ago

Create some integrations tests based on the representative sample of NIFS proposal ID values (see issue #10 for details)

Nat1405 commented 4 years ago

Just an updated on this: I've reduced 10 of the 35 representative datasets by hand so far as a preliminary test (I'm not sure if we'll be able to make automatic integration tests on cavern before we figure out batch but maybe Brian has some ideas). All 10 of them stopped before completing because of either a bug in Nifty4Gemini or insufficient calibrations being present. @dbohlender @ijiraq do you mind if I spend a bit of my time fixing Nifty4Gemini bugs so these data reductions can succeed without manual intervention?

Nat1405 commented 4 years ago

(Nifty4Gemini works very well for the programs it was initially built for but I don't think it's ever been tested at this scale).

dbohlender commented 4 years ago

Yes, I'd say this should be a pretty high priority. Yesterday you commented on the poor registration in x/y of two merged cubes. Is that also something that you are able to look at fixing?

Nat1405 commented 4 years ago

Thank you.

Yes! I'll definitely look into that too. Hopefully there's an easy headers solution for that one. I'm thinking I'll track the various Nifty4Gemini bug fixes in the Nifty4Gemini issues tab if that sounds good to you.

dbohlender commented 4 years ago

👍 (Can you tell I read a bit about GitHub yesterday? I know how to send emoji's now.)

Nat1405 commented 4 years ago

This Nifty4Gemini PR addresses this. Integration tests are ready for a basic review; I've only added the bare minimum to be happy the reductions are working well.

https://github.com/Nat1405/Nifty4Gemini/pull/26