search

mdabros / SharpLearning

Machine learning for C# .Net
MIT License
384 stars 85 forks source link

0.31.6.0: Adds implicit and explicit conversions from double[][] to F64Matrix #124

Closed jameschch closed 5 years ago

jameschch commented 5 years ago

This is a simple way to address the need to expose an API surface that accepts double[][] rather than an F64Matrix. Instead of changing all interfaces and implementations, I have made double[][] implicitly convertible to F64Matrix. This can be considered a convenience and a temporary workaround for #20 and #115 .

mdabros commented 5 years ago

@jameschch I think this is a good idea. Even though it is a bit memory heavy to convert between double[][] and F64Matrix in this way. But overall, it makes life easier for developers that work with manageable data sizes, and usually use the jagged array type.

Maybe you could add similar methods for supporting multidim arrays double[,]? It can be in a separate/later pull request if that suits you better.

jameschch commented 5 years ago

Build seems to be failing after I added some line breaks.

mdabros commented 5 years ago

@jameschch Failing tests seems to be unrelated to your changes, so the failing build is most likely caused by some unstable unit tests. I will have a look at it later this week.

mdabros commented 5 years ago

@jameschch Since the checks fails because of flaky unit tests, I think we should just merge this, and I can look into the flaky tests in another PR.

If you increment the version from 0.31.5.0 to 0.31.6.0 I can merge this. You might need to merge with the master first since #88 was completed in the meantime.

mdabros commented 5 years ago

@jameschch I have completed the pull request. Thanks for the contribution! packages should be up shortly.

best regards Mads