As we build more and more functionality into the library, it would be a good idea to make sure that we don't do anything that inadvertently cripples the efficiency. I propose the following:
Write a selection of benchmarks using tasty-bench as a test suite (these will also be able to serve as integration tests!)
If the commit is tagged write these results to a .csv file, and push them into a bench-data branch under tag.csv
On push, if there is a previous tag on this branch (or main) pull down the csv file for that tag, and use it as a comparison, allowing us to test that things don't deviate too much.
As we build more and more functionality into the library, it would be a good idea to make sure that we don't do anything that inadvertently cripples the efficiency. I propose the following:
tasty-benchas a test suite (these will also be able to serve as integration tests!).csvfile, and push them into abench-databranch undertag.csvmain) pull down the csv file for that tag, and use it as a comparison, allowing us to test that things don't deviate too much.Some resources on this: