cache warmer

[benmosher]

Should theoretically cut time by 25% or so by capturing exports while linting; this avoids the second parse operation for files that have already been linted that are then referenced from others.

• [ ] profile as % of total rule execution time:
  • file load time: ___
  • parse time: ___
  • export map building: ___

might be simpler to just move toward an FS cache implementation of the export map since that will be needed anyway

[benmosher]

For the tests, out of 983ms:

read: 3.450444ms parse: 99.025113ms full: 159.063341ms

So reading is inconsequential, and parsing takes 10% of the time, and export collection is 15%.

This is probably not a great sample, but does imply that parsing is a decent chunk of the work.

[benmosher]

nvm on this, memo-parser is a more complete solution