692806 - Document/CSS cannot use SVG filter when url points to data: URI

Status: RESOLVED DUPLICATE of bug 686013

(Core :: SVG, defect)

Description

[Michael Donning]

Attachment: Testcase demonstrating the use of svg via data URIs in css

User Agent: Mozilla/5.0 (Windows NT 5.1; rv:7.0.1) Gecko/20100101 Firefox/7.0.1
Build ID: 20110928134238

Steps to reproduce:

I tried to apply an SVG effect using a CSS style. The source url for the svg was a data URI. (Fragments in data URIs are supported starting from Firefox 6.0)

https://developer.mozilla.org/En/Applying_SVG_effects_to_HTML_content

Additionally i used a SVG data URI in CSS for definining a background-image.


Actual results:

The html element(s) using the style were invisible. Additionally there was a security warning on the error console noting the different origins of the document and the data uri.

The CSS SVG background-image rendered fine and produced no message on the error console.


Expected results:

This seems inconsistent to me. Either both or none of the SVG data URIs should be accepted.

Enabling the use of data URIs for svg filters in CSS would also (probably) enable the possibility to use them in userContent.css.

Note:
I was asked to file this new BUG in a reply to my comment in BUG 308590 which i misunderstood as the same problem.
https://bugzilla.mozilla.org/show_bug.cgi?id=308590

Comment 1

[Michael Donning]

Comment on attachment 565553 [details]
Testcase demonstrating the use of svg via data URIs in css

This file shows three paragraphs:
#1: The use of a css SVG filter (gaussian blur) via embedded SVG
#2: (Invisible) The same SVG filter as in #1 but via data URI
#3: Successful use of svg data URIs for a css background-image.

The error console should list a single security error concerning the (filter) data URI (#2) only.

Comment 2

[Daniel Holbert [:dholbert]]

Attached testcase is fixed by my patch for bug 686013.  Marking as a dupe -- same underlying issue. (overzealous same-origin restrictions when referencing a fragment of a data URI)

Thanks for filing -- I'll be sure to include a regression test based off of the testcase here when I check in the patch for that bug!