You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There are two (probably) unrelated things in the driver's performance that I'd like to understand:
Why do very simple AST passes take so long? E.g., for drivers containing derivers, why do the AST passes turning off the warnings w32 and w60 take so long?
- [ ] Look for a non-IP sensitive example where those passes take long.
- [ ] Rewrite them using compiler-lib's AST traverse workflow via records.
- [ ] Compare the two timing results: standard Ppxlib traverse workflow via classes vs compiler-lib's records workflow. Is there a significant difference?
In some very dubious cases of PPXs, the number of AST passes varies strongly depending on which syntax is used in the AST the PPX is run on. How can that happen?
- [ ] Reproduce this.
- [ ] Find an explanation to find out whether that's dubious PPX behavior or a Ppxlib performance bug.
The text was updated successfully, but these errors were encountered:
There are two (probably) unrelated things in the driver's performance that I'd like to understand:
- [ ] Look for a non-IP sensitive example where those passes take long.
- [ ] Rewrite them using
compiler-lib
's AST traverse workflow via records.- [ ] Compare the two timing results: standard Ppxlib traverse workflow via classes vs
compiler-lib
's records workflow. Is there a significant difference?- [ ] Reproduce this.
- [ ] Find an explanation to find out whether that's dubious PPX behavior or a Ppxlib performance bug.
The text was updated successfully, but these errors were encountered: