Standalone GQ: Output NaNs when exception is caught

[WardBrian]

Submission Checklist

• Run unit tests: ./runTests.py src/test/unit
• Run cpplint: make cpplint
• Declare copyright holder and open-source license: see below

Summary

Closes stan-dev/cmdstan#1007. Since stan-dev/stanc3#1165 we have made sure that write_array always populates the array with NaNs, so writing it out is safe even after an exception.

Intended Effect

Standalone generate quantities behavior now matches that of sampling (nans output if write_array fails) and the number of rows in the output of standalone generated quantities is now always the number of rows in the input fitted_params.

How to Verify

The example from stan-dev/cmdstan#1007 works well

Side Effects

Documentation

Copyright and Licensing

Please list the copyright holder for the work you are submitting (this will be you or your assignee, such as a university or company):

Simons Foundation

By submitting this pull request, the copyright holder is agreeing to license the submitted work under the following licenses:

• Code: BSD 3-clause (https://opensource.org/licenses/BSD-3-Clause)
• Documentation: CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/)

[bob-carpenter]

Looks good other than scope of a logger in one file.

[bob-carpenter]

If this isn't going to return, then the following logger_ call outside the catch duplicates the ss message. The full context is this:

    try {
      model.write_array(rng, const_cast<std::vector<double>&>(draw), params_i,
                        values, false, true, &ss);
    } catch (const std::exception& e) {
      if (ss.str().length() > 0)
        logger_.info(ss);
      logger_.info(e.what());
    }
    if (ss.str().length() > 0)
      logger_.info(ss);

The conditional logging after the catch should be moved into the try block to avoid duplicating logger_.info(ss).

This is also making me think the logger itself should be ignoring empty string messages rather than duplicating this everywhere.

[WardBrian]

Good catch.

Ignoring empty strings has a (nearly 5 year old!) issue open: #2577

[bob-carpenter]

Thanks for adding the test.

[bob-carpenter]

Thanks for the update.

[bob-carpenter]

Thanks, and feel free to merge when tests finish.