Work log 2016-08-09

Oops, benchmarks weren’t getting their stding, which is why annotating was failing.

Clustering is still a memory hog, so I’ve bumped the RAM limit to 5GB.

IsaCoSy is a bit of a mess. Example doesn’t work for a fresh theory, so I need to find the appropriate function to call. Probably Constraints.* or SynthNames.* or something.

TIP won’t annotate since its environment doesn’t contain tip-benchmark-sig. Tests were making their own env; switched to explore.extractedEnv instead.

Tests keep failing due to lack of available workers. Annoying consequence of using nix-build inside the build command. Would be nice to avoid it, but looks like too much effort ATM.

Maybe multipl-output derivations might help? THat way we can still reference things like rawAnnotated.stdout, but they won’t be built immediately.

Trying it on dumps to begin with; if it works without much pain try it on annotated, then clustered, etc. In the process, move tests to testDrvs to avoid nix-build.