Bug 167515 – Round CPU Usage diagnostic logging to 2 significant digits

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit2

[Chris Dumez]

Round CPU Usage diagnostic logging to 2 significant digits.

[Chris Dumez]

<rdar://problem/30236297>

[Chris Dumez]

Created attachment 299943 [details]
Patch

[Antti Koivisto]

Comment on attachment 299943 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=299943&action=review

> Source/WebKit2/UIProcess/PerActivityStateCPUUsageSampler.cpp:80
> +static String toStringRoundingSignificantFigures(double value, unsigned significantFigures)
> +{
> +    DecimalNumber decimal(value, RoundingSignificantFigures, significantFigures);
> +    NumberToLStringBuffer buffer;
> +    unsigned length = decimal.toStringDecimal(buffer, WTF::NumberToStringBufferLength);
> +    return String(buffer, length);
> +}

Is this somehow better than String::number(cpuUsage, 2)?

[Chris Dumez]

Comment on attachment 299943 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=299943&action=review

>> Source/WebKit2/UIProcess/PerActivityStateCPUUsageSampler.cpp:80
>> +}
> 
> Is this somehow better than String::number(cpuUsage, 2)?

String::number() takes a number of "decimal digits". I really need a number of "significant figures". 10 has 1 significant figure. 10.2 has 3 significant figures.

[WebKit Commit Bot]

Comment on attachment 299943 [details]
Patch

Clearing flags on attachment: 299943

Committed r211295: <http://trac.webkit.org/changeset/211295>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Antti Koivisto]

> I really need a number
> of "significant figures". 10 has 1 significant figure. 10.2 has 3
> significant figures.

I thought that is what String::number() with precision argument does.

[Chris Dumez]

(In reply to comment #7)
> > I really need a number
> > of "significant figures". 10 has 1 significant figure. 10.2 has 3
> > significant figures.
> 
> I thought that is what String::number() with precision argument does.

Oh, I'll double check then. Using String::number() would be nicer indeed if it does what I want.