Based off of our discussion yesterday, this PR tests saving a pruned version of the kb to the cache. On some functions (tar_dirname), this changed runtime from 33s to 9s, but on others, (parse_branch), this changed runtime from 1m46s to 2m45s. I'm not really sure if this is the approach we want to go for.
This also might fix #247. I wasn't able to reproduce this bug unfortunately, but I had to update some parts of the architecture retrieval API. If you could test this for me, that would be great!
Based off of our discussion yesterday, this PR tests saving a pruned version of the kb to the cache. On some functions (tar_dirname), this changed runtime from 33s to 9s, but on others, (parse_branch), this changed runtime from 1m46s to 2m45s. I'm not really sure if this is the approach we want to go for.
This also might fix #247. I wasn't able to reproduce this bug unfortunately, but I had to update some parts of the architecture retrieval API. If you could test this for me, that would be great!