-
Notifications
You must be signed in to change notification settings - Fork 184
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Optimize sourcemap processing #1617
base: master
Are you sure you want to change the base?
Conversation
25c2733
to
9218fa3
Compare
I can’t seem to reproduce the CI failure:
This test builds fine on my machine. Edit: never mind, it was an off-by-one error in an assertion. |
7ca0b47
to
ac75d97
Compare
One test is failing because reading “index maps” (composite source maps) is not supported for now—something that is not required for js_of_ocaml to function. I will fix this test soon. |
ab355a0
to
64f5df3
Compare
I just tested this change on an incremental build for one of our largest executables and am getting this timing for Pre: 2.469s Very cool! |
ec3ead0
to
3ffa259
Compare
This is ready for review. I don’t quite understand the build problems on the 5.1 CI, but they don’t seem related to this PR. |
01f15f7
to
fb76558
Compare
My bad, it actually was due to a missing |
bfbccf2
to
3b5260a
Compare
3b5260a
to
ae4c709
Compare
This improves the efficiency of sourcemap processing, chiefly during linking, but also in serializing and deserializing sourcemaps.
Currently, when linking with source maps enabled, js_of_ocaml will decode the source maps for each input file completely into an in-memory explicit mapping, then merge them, then re-encode the merged sourcemap. The merging algorithm involves superlinear sorting operations. However, the format of sourcemaps allows to avoid sorting for the concatenation of generated files.
In addition, all operations performed by linking (concatenation of files, removal and addition of lines) can be performed directly on the encoded form, which further avoids a full parsing to an in-memory representation.
When compiling js_of_ocaml with itself, the final linking step is accelerated by a speedup of about 2.3x with OCaml 5.2.0+fp:
fix #1446