Phil Miller - NOAA

One easy way would be to just change what unit the BMI input variable advertises, to match the scaling. Everything else should just do the right thing then, I think?...

*Original date: 2015-12-02 18:05:16* --- I just tried adding some asserts to the message delivery path that the envelope has the `srcPe` fields set correctly. They don't fire under testing....

*Original date: 2017-09-25 21:53:47* --- I'll try to take care of this quickly.

*Original date: 2017-09-26 17:23:32* --- https://charm.cs.illinois.edu/gerrit/3079

*Original date: 2017-09-27 20:04:25* --- The straight-forward fix for this messes up AMPI intercommunicator creation. Deferring for now, since the full fix will require some deeper surgery.

Does it really make sense to call this `deep_copy`? As I understand, those are generally not supposed to be doing any computation. An arbitrary `operator=` that could be called between...

Could we get a coherent treatment of `scalar` and `none` in a single proposal, rather than split between this and #184? For instance, why can't we just elaborate the specification...

This is something that would only really be appropriate for a new major version of the standard, because it would make previously-conformant models that return `scalar` for some variables non-conformant.

After preprocessing the file to a self-contained test case with `-E`, I can reproduce the failure with just `icpc -xCORE-AVX2 -g -fPIE -std=c++14 -c normals.cpp`. When I drop the detailed...

Just ` icpc -xCORE-AVX2 -c normals.cpp` fails, so there's something wrong in optimization or code generation