Bug 234001 – [css-contain][Performance test] Add test contain-paint-text-nowrap.html

• Component: CSS
• Status: RESOLVED FIXED

[Rob Buis]

Add test contain-paint-text-nowrap.html.

[Rob Buis]

Created attachment 446355 [details]
Patch

[Rob Buis]

Test results on my M1 mbp:

Containment on:
:Time -> [17, 17, 14, 15, 14, 16, 17, 17, 16, 17, 17, 16, 17.000000000000057, 16.999999999999943, 14, 17, 17.000000000000057, 15.999999999999943, 16, 17] ms
    mean: 16.2 ms
    median: 16.99999999999997 ms
    stdev: 1.1050125029061675 ms
    min: 14 ms
    max: 17.000000000000057 ms

Containment off:
:Time -> [26, 23, 24, 23, 20, 21, 21, 21, 20, 21, 21, 19, 19, 20, 19, 19, 19, 19, 19, 20] ms
    mean: 20.7 ms
    median: 20 ms
    stdev: 1.949358868961793 ms
    min: 19 ms
    max: 26 ms

[Rob Buis]

Test results on my M1 mbp:

Containment on:
:Time -> [17, 17, 14, 15, 14, 16, 17, 17, 16, 17, 17, 16, 17.000000000000057, 16.999999999999943, 14, 17, 17.000000000000057, 15.999999999999943, 16, 17] ms
    mean: 16.2 ms
    median: 16.99999999999997 ms
    stdev: 1.1050125029061675 ms
    min: 14 ms
    max: 17.000000000000057 ms

Containment off:
:Time -> [26, 23, 24, 23, 20, 21, 21, 21, 20, 21, 21, 19, 19, 20, 19, 19, 19, 19, 19, 20] ms
    mean: 20.7 ms
    median: 20 ms
    stdev: 1.949358868961793 ms
    min: 19 ms
    max: 26 ms

[Radar WebKit Bug Importer]

<rdar://problem/86519206>

[Simon Fraser (smfr)]

Comment on attachment 446355 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=446355&action=review

> PerformanceTests/ChangeLog:13
> +        * Paint/contain-paint-nowrap.html: Added.

Might be better to have a "containment" directory, rather than a "Paint" directory.

> PerformanceTests/Paint/contain-paint-nowrap.html:40
> +            document.body.getBoundingClientRect();

Why the forced layout?

> PerformanceTests/Paint/contain-paint-nowrap.html:44
> +        var isDone = false;

var -> let?

> PerformanceTests/Paint/contain-paint-nowrap.html:52
> +        var startTime;
> +        var height = 0;

var -> let?

[Simon Fraser (smfr)]

(In reply to Rob Buis from comment #3)
> Test results on my M1 mbp:
> 
> Containment on:
> :Time -> [17, 17, 14, 15, 14, 16, 17, 17, 16, 17, 17, 16,
> 17.000000000000057, 16.999999999999943, 14, 17, 17.000000000000057,
> 15.999999999999943, 16, 17] ms
>     mean: 16.2 ms
>     median: 16.99999999999997 ms
>     stdev: 1.1050125029061675 ms
>     min: 14 ms
>     max: 17.000000000000057 ms
> 
> Containment off:
> :Time -> [26, 23, 24, 23, 20, 21, 21, 21, 20, 21, 21, 19, 19, 20, 19, 19,
> 19, 19, 19, 20] ms
>     mean: 20.7 ms
>     median: 20 ms
>     stdev: 1.949358868961793 ms
>     min: 19 ms
>     max: 26 ms

Please do some rounding of the output: one decimal place for milliseconds should be enough.

[Rob Buis]

Created attachment 447438 [details]
Patch

[Rob Buis]

Created attachment 447440 [details]
Patch

[Rob Buis]

Comment on attachment 446355 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=446355&action=review

>> PerformanceTests/ChangeLog:13
>> +        * Paint/contain-paint-nowrap.html: Added.
> 
> Might be better to have a "containment" directory, rather than a "Paint" directory.

Yes, containment directory is a good idea, I added one, and moved the existing containment tests there too.

>> PerformanceTests/Paint/contain-paint-nowrap.html:40
>> +            document.body.getBoundingClientRect();
> 
> Why the forced layout?

Does not seem needed indeed.

>> PerformanceTests/Paint/contain-paint-nowrap.html:44
>> +        var isDone = false;
> 
> var -> let?

Done.

>> PerformanceTests/Paint/contain-paint-nowrap.html:52
>> +        var height = 0;
> 
> var -> let?

Done.

[EWS]

Committed r287179 (245348@main): <https://commits.webkit.org/245348@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 447440 [details].

[Rob Buis]

(In reply to Simon Fraser (smfr) from comment #6)
> (In reply to Rob Buis from comment #3)
> > Test results on my M1 mbp:
> > 
> > Containment on:
> > :Time -> [17, 17, 14, 15, 14, 16, 17, 17, 16, 17, 17, 16,
> > 17.000000000000057, 16.999999999999943, 14, 17, 17.000000000000057,
> > 15.999999999999943, 16, 17] ms
> >     mean: 16.2 ms
> >     median: 16.99999999999997 ms
> >     stdev: 1.1050125029061675 ms
> >     min: 14 ms
> >     max: 17.000000000000057 ms
> > 
> > Containment off:
> > :Time -> [26, 23, 24, 23, 20, 21, 21, 21, 20, 21, 21, 19, 19, 20, 19, 19,
> > 19, 19, 19, 20] ms
> >     mean: 20.7 ms
> >     median: 20 ms
> >     stdev: 1.949358868961793 ms
> >     min: 19 ms
> >     max: 26 ms
> 
> Please do some rounding of the output: one decimal place for milliseconds
> should be enough.

AFAIK this is global behaviour of the perf test framework, or is there a way to fine-tune the accuracy per test through some parameter?

[Simon Fraser (smfr)]

Probably not. We should fix the framework!