Code Refactor - Disentangle domain setup and run configuration to support domain caching#104
Code Refactor - Disentangle domain setup and run configuration to support domain caching#104vedran-kasalica wants to merge 5 commits into
Conversation
|
Hi @vedran-kasalica, what is the plan for this PR? |
Hi @eladrion, this was my effort to disentangle the encodings of the We talked about caching once, and this could be a nice student project. It could speed up APE drastically. *I updated the PR description to reflect that. |
|
Hi @vedran-kasalica. Okay, get that. I will discuss that internally at UP and this PR will then stay open for now |
thanks! I opened an issue as well #128 |
Caching of the domain annotations could speed up APE drastically.
This was my effort to disentangle the encodings of the domain annotations and the run configuration, as it is crucial for caching the domain annotations. It is motivated by the fact that often domain encoding (into cnf) takes longer than the problem solving, and considering that our domains are semi-static, if we cache the domian encoding (cache the cnf encoding of the domain per each length, because each workflow length has different solution space), we can just re-use it and append the run configuration (run configuration would be cnf encoding of workflow inputs, outputs, constraints, etc.). However, I didn't get to finish the refactor, that's why it's still a draft.
Related issue #128