feat: Add ingestion of matrices

[joakimlinde]

No description provided.

[joakimlinde]

Looking at R API issue... Fixed.

[krlmlr]

Thanks. Is there a good way to make this feature opt-in, based on ConvertOpts ?

I see this adds some support for complex numbers. Could this have been a separate PR? Same question regarding opt-in behavior.

[krlmlr]

Should this be DATAPTR_RO() here and elsewhere?

[joakimlinde]

Yes. Fixed.

[krlmlr]

Should this check and the while loop be swapped?

[joakimlinde]

Sure, I can swap them. You mean that if I have a matrix inside a structure it would work better? Not sure if I've thought about that case... Is that a case we want to support?

[krlmlr]

For completeness and accuracy, yes. Not that I expect it to occur; if it does nevertheless, we'll save a lot of time later by doing a small step now.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 87.05882% with 11 lines in your changes missing coverage. Please review.

Project coverage is 52.62%. Comparing base (5c88c6f) to head (a2b189e).
Report is 1539 commits behind head on main.

Files with missing lines	Patch %	Lines
src/scan.cpp	83.33%	8 Missing ⚠️
src/types.cpp	90.90%	3 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #1150       +/-   ##
===========================================
- Coverage   63.61%   52.62%   -10.99%     
===========================================
  Files         123      146       +23     
  Lines        5266    10025     +4759     
===========================================
+ Hits         3350     5276     +1926     
- Misses       1916     4749     +2833     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[joakimlinde]

Is there a good way to make this feature opt-in, based on ConvertOpts ?

No, I don't think so. Here is why. If the user gives me a matrix I only have two options. Either reject it as "I don't know what to do with a matrix" or slice up the matrix into arrays and push that into the database. The intent is clear from the user - he/she is giving me a matrix and there is only one constructive thing I can do with it. If the user had given me lists instead it would have been different. In that case I would have to know what to do with them, either create lists or arrays in the database. In that case the user would have to give me additional direction through options.

I see this adds some support for complex numbers. Could this have been a separate PR? Same question regarding opt-in behavior.

I added support so I can generate an error message saying that complex numbers are not supported. I could do it in a different way by looking at the SEXP type instead if that works better for you. It would break the pattern in scan.cpp though — look a little bit as a quick-and-dirty.

[joakimlinde]

The Ubuntu-22.04 (4.2) check fails. Seems like the snapshot files didn't make it across to Ubuntu. Can I ask you to manually rerun the test?

UPDATE: After triggering a few reruns (using dummy changes) the tests are now passing.

[krlmlr]

Thanks!