Bug 150101 – Web Inspector: CSS Data URIs count against page weight twice

• Status: RESOLVED
• Resolution: FIXED
• Component: Web Inspector

[Lanny Heidbreder]

Created attachment 263021 [details]
Example HTML file that shows the problem

OVERVIEW:

This is a repeat of bug 75555, which was closed (automatically?) when the new Web Inspector appeared.

Any CSS data URI is counted twice against page weight: once for the CSS code, and once for the separate resource entry for the image or font represented by the data URI.

STEPS TO REPRODUCE:

Open Web Inspector, load a single HTML page that uses a data URI in a <style> element, and look at the Page Weight indicator.

EXPECTED RESULTS:

The Page Weight total shows the exact HTML file size.

ACTUAL RESULTS:

The Page Weight total shows the sum of the HTML file size, PLUS the size of the image or font represented by the data URI, since it gets a separate Resource entry

ADDITIONAL INFO:

You can see this in action at http://75thtrombone.com/links/webkit/page-weight-test.html . That page is also attached.

[Radar WebKit Bug Importer]

<rdar://problem/23096774>

[Timothy Hatcher]

Thanks for the bug report!

Should be a simple fix on DefaultDashboard's _resourceSizeDidChange to ignore data URLs.

[Matt Baker]

Created attachment 263853 [details]
[Patch] Proposed Fix

[Matt Baker]

I considered using Resource.urlComponents.scheme, but the data URI resource's scheme was null (expected "data").

[Joseph Pecoraro]

(In reply to comment #4)
> I considered using Resource.urlComponents.scheme, but the data URI
> resource's scheme was null (expected "data").

Maybe parseURL() doesn't properly handle a data URI. I think I was running into this recently.

[Matt Baker]

(In reply to comment #5)
> (In reply to comment #4)
> > I considered using Resource.urlComponents.scheme, but the data URI
> > resource's scheme was null (expected "data").
> 
> Maybe parseURL() doesn't properly handle a data URI. I think I was running
> into this recently.

I'm certain the scheme property was null at least once while refreshing the attached example page, but I can't reproduce the issue now. Going to switch to checking the scheme field since we do this elsewhere (CSSStyleDeclarationSection.js:516).

[Matt Baker]

Created attachment 263866 [details]
[Patch] Proposed Fix

[WebKit Commit Bot]

Comment on attachment 263866 [details]
[Patch] Proposed Fix

Clearing flags on attachment: 263866

Committed r191486: <http://trac.webkit.org/changeset/191486>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.