-
Notifications
You must be signed in to change notification settings - Fork 413
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
[performance] Zero-build regression 3.11 - 3.14 #10088
Comments
Do you want put up some |
Indeed it seems that the
|
I don't know enough
Code in question is here dune/src/dune_rules/coq/coq_rules.ml Lines 521 to 559 in b5a4f6a
|
Regression from ocaml#9231 Fixes ocaml#10088 Signed-off-by: Emilio Jesus Gallego Arias <[email protected]>
Regression from ocaml#9231 Fixes ocaml#10088
Indeed, after memoizing the map construction, the zero build time goes from 5 sec to 0.3 , will submit a PR shortly. |
Regression from ocaml#9231 Fixes ocaml#10088 Signed-off-by: Emilio Jesus Gallego Arias <[email protected]>
Regression from #9231 Fixes #10088 Signed-off-by: Emilio Jesus Gallego Arias <[email protected]> Co-authored-by: Emilio Jesus Gallego Arias <[email protected]>
CHANGES: ### Fixed - When a directory is changed to a file, correctly remove it in subsequent `dune build` runs. (ocaml/dune#9327, fix ocaml/dune#6575, @emillon) - Fix a problem with the doc-new target where transitive dependencies were missed during compile. This leads to missing expansions in the output docs. (ocaml/dune#9955, @jonludlam) - coq: fix performance regression in coqdep unescaping (ocaml/dune#10115, fixes ocaml/dune#10088, @ejgallego, thanks to Dan Christensen for the report) - coq: memoize coqdep parsing, this will reduce build times for Coq users, in particular for those with many .v files (ocaml/dune#10116, @ejgallego, see also ocaml/dune#10088) - on Windows, use an unicode-aware version of `CreateProcess` to avoid crashes when paths contains non-ascii characters. (ocaml/dune#10212, fixes ocaml/dune#10180, @emillon)
Hi folks,
I'm forwarding this perf regression from Coq users, as it seems interesting.
The relevant bit seems this zero-build data:
That's almost 2 seconds so worth having a look IMHO.
Note that we identified a much worse performance regression while investigating this issue, see #10116
The text was updated successfully, but these errors were encountered: