A warning about missing pandoc-tangle causes spurious errors about evm-semantics

[pirapira]

$ java -version
java version "1.8.0_144"
Java(TM) SE Runtime Environment (build 1.8.0_144-b01)
Java HotSpot(TM) 64-Bit Server VM (build 25.144-b01, mixed mode)

[ehildenb]

Confirmed (though the test is passing, there is an enormous amount of output).

Better to reproduce with ./tests/ci/with-k uiuck ./Build test tests/proofs/hkg/approve-spec.k (because ./Build tests quick selects tests randomly).

The enormous amount of output is more an issue directly with K, I'll open an issue there so we can start getting more informative messages out of K when proofs aren't going perfectly.

[ehildenb]

And your intuition is correct that pandoc-tangle missing is not causing this issue, though that may be why the test reports as failed when on a system with pandoc-tangle it will report as passed. I'll investigate that too. The repository https://github.com/ehildenb/pandoc-tangle has a Linux binary for pandoc-tangle. I can see about getting a MacOS binary in there too if you need that.

[ehildenb]

Confirmed that when pandoc-tangle binary is not present, reports test failure even though test passes because the message === WARNING: pandoc-tangle not installed ... is being interpreted as output and compared to the value true, which is the expected output.

[pirapira]

Then this issue is more like "warning about missing pandoc-tangle causes spurious errors." I'll change the title.

[ehildenb]

On branch quieter-proofs, you should be able to (replace bash with your preferred shell)

./tests/ci/with-k uiuck bash
./Build tests proofs approve-spec

It will output all the K warnings still, but should now say passed instead of failed.

I'll wait to turn this into a PR until we have K being quieter about the translation issues, and bump the submodule version when that happens.

[pirapira]

@ehildenb I can confirm the behavior on quieter-proofs.

[pirapira]

This issue has disappeared.