Bug 45576 – fast/files/apply-blob-url-to-img.html timeout on Leopard

• Component: WebCore JavaScript
• Status: RESOLVED FIXED

[Jian Li]

fast/files/apply-blob-url-to-img.html timeout on Leopard.

[Jian Li]

Created attachment 67267 [details]
Proposed Patch

[David Levin]

Comment on attachment 67267 [details]
Proposed Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=67267&action=prettypatch

> WebKit/chromium/ChangeLog:8
> +        * src/ResourceHandle.cpp: Remove the unneeded static comment.
s/unneeded/no longer applicable/

[Jian Li]

Committed as http://trac.webkit.org/changeset/67503.

[WebKit Review Bot]

http://trac.webkit.org/changeset/67503 might have broken GTK Linux 32-bit Release

[Alexey Proskuryakov]

I don't think that this is a good fix. supportsBufferedData is a check for platform capabilities, and adding a check that internal to WebCore on top of it is highly confusing. I think that a different fix is needed here.

Also, I may not understand the purpose of BlobResourceHandle, but it seems to share the same problem - although ResourceHandle is supposed to be a platform abstraction, it now has a subclass that contains WebCore logic.

[Alexey Proskuryakov]

Since this also broke build, rolling out seems like an appropriate course of action to me.

[Alexey Proskuryakov]

This change didn't unskip the test. Done in <http://trac.webkit.org/changeset/107879>.

It is disappointing that the above comments have not be acted upon.

[Jian Li]

(In reply to comment #7)
> This change didn't unskip the test. Done in <http://trac.webkit.org/changeset/107879>.
> 
> It is disappointing that the above comments have not be acted upon.

My apology for not addressing the concern quickly. We did go through quite a lot discussions on which layer we wanted to hook up with the URL request processing. I agree that BlobResourceHandle inherited from ResourceHandle might not be a ideal solution to process the blob URL request. But the discussions did not lead to any other better choice at that time.