1797690 - Cover source-map npm package performance on DAMP

Status: RESOLVED FIXED

(DevTools :: Debugger, task)

Description

[Alexandre Poirot [:ochameau]]

The source-map package is perf sensitive as it impact the speed of the debugger to process sourcemapped files, but also impact external projects like babel, webpack, typescript,.. which are still using our npm package (some started using forks because of performance reason).

Bug 1797274 focused on testing the internal layer within Gecko/mozilla-central which wraps the source-map npm package and do more.
This current bug will focus on the npm package API only.

Comment 1

[Alexandre Poirot [:ochameau]]

This depends on bug 1794693 as today, the source-map package isn't exposed in mozilla-central.
Instead this is wrapped into a webpack bundle and we can't access to raw source-map package modules.
https://searchfox.org/mozilla-central/rev/2d1097c27f31cb15f7588b043d691cda9710d14b/devtools/client/shared/source-map/index.js

Comment 2

[Alexandre Poirot [:ochameau]]

Attachment: Bug 1797690 - [devtools] Track performance of source-map npm package.

Comment 3

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/244069f551ca
[devtools] Track performance of source-map npm package. r=bomsy,jdescottes,perftest-reviewers,sparky

Comment 4

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/244069f551ca

Comment 5

[Julian Descottes [:jdescottes]]

== Change summary for alert #36028 (as of Thu, 10 Nov 2022 09:33:48 GMT) ==

Regressions:

Ratio	Test	Platform	Options	Absolute values (old vs new)
40%	damp source-map-loader.init.DAMP	windows10-64-shippable-qr	e10s fission stylo webrender-sw	71.72 -> 100.47
39%	damp source-map-loader.init.DAMP	windows10-64-shippable-qr	e10s fission stylo webrender	72.34 -> 100.22
36%	damp source-map-loader.init.DAMP	linux1804-64-shippable-qr	e10s fission stylo webrender-sw	102.12 -> 139.32
35%	damp source-map-loader.init.DAMP	linux1804-64-shippable-qr	e10s fission stylo webrender	104.01 -> 140.35
30%	damp source-map-loader.init.DAMP	macosx1015-64-shippable-qr	e10s fission stylo webrender	65.34 -> 84.88
27%	damp source-map-loader.init.DAMP	macosx1015-64-shippable-qr	e10s fission stylo webrender-sw	65.72 -> 83.72
24%	damp complicated.styleeditor.open.DAMP	windows10-64-shippable-qr	e10s fission stylo webrender	323.22 -> 401.55
23%	damp complicated.styleeditor.open.DAMP	windows10-64-shippable-qr	e10s fission stylo webrender-sw	330.63 -> 407.18
22%	damp complicated.styleeditor.open.DAMP	linux1804-64-shippable-qr	e10s fission stylo webrender-sw	350.48 -> 426.70
18%	damp complicated.styleeditor.open.DAMP	macosx1015-64-shippable-qr	e10s fission stylo webrender-sw	290.17 -> 343.59
...	...	...	...	...
3%	damp simple.jsdebugger.open.DAMP	linux1804-64-shippable-qr	e10s fission stylo webrender	489.16 -> 503.29
3%	damp custom.jsdebugger.open.DAMP	linux1804-64-shippable-qr	e10s fission stylo webrender	692.82 -> 712.53
3%	damp simple.netmonitor.requestsFinished.DAMP	windows10-64-shippable-qr	e10s fission stylo webrender	101.64 -> 104.48
3%	damp browser-toolbox.debugger-ready.DAMP	macosx1015-64-shippable-qr	e10s fission stylo webrender	998.07 -> 1,024.60
2%	damp browser-toolbox.debugger-ready.DAMP	macosx1015-64-shippable-qr	e10s fission stylo webrender-sw	995.31 -> 1,018.13

Improvements:

Ratio	Test	Platform	Options	Absolute values (old vs new)
6%	damp simple.netmonitor.close.DAMP	windows10-64-shippable-qr	e10s fission stylo webrender	9.41 -> 8.81
5%	damp source-map.eachMapping.DAMP	linux1804-64-shippable-qr	e10s fission stylo webrender	246.80 -> 234.57
5%	damp complicated.jsdebugger.open.DAMP	macosx1015-64-shippable-qr	e10s fission stylo webrender-sw	979.89 -> 933.54
4%	damp source-map.originalPositionFor.DAMP	linux1804-64-shippable-qr	e10s fission stylo webrender	276.08 -> 265.16
4%	damp source-map-loader.getOriginalLocation.DAMP	linux1804-64-shippable-qr	e10s fission stylo webrender-sw	123.90 -> 119.38
2%	damp custom.netmonitor.open.DAMP	macosx1015-64-shippable-qr	e10s fission stylo webrender-sw	178.98 -> 175.30

For up to date results, see: https://treeherder.mozilla.org/perfherder/alerts?id=36028

Comment 6

[Alexandre Poirot [:ochameau]]

The regression were expected.

It relates to moving from bundle (one big file) to independant modules (one file per module).
This was necessary to stop having the manual webpack build step and will later help migrate to ES Modules.
Hopefully we will haver better performance when moving to ESM. Today modules are loaded via some JS-implemented worker loader.