scrolling off the side of http://bugzilla.opendarwin.org/attachment.cgi?id=6042 shows the issue
Created attachment 12159 [details] Reduced testcase. The problem is actually, because of <tspan> in <text>, that breaks "gradient-on-text".
I can't tell if this is fixed. Niko?
(In reply to comment #2) > I can't tell if this is fixed. Niko? > Nope, problem is still persistent. Not sure how to fix yet. Niko
Created attachment 16187 [details] Initial patch Not ready for review yet. Just for Oliver :-)
Created attachment 16188 [details] Updated patch The reduced testcase is obsolete as the problem gets highlighted by a different (already commited) batik/text testcase.
Created attachment 16189 [details] Updated patch v2 Some last hot fixes - demanded by Oliver. Seperated LayoutTests/ and patch upon request.
Created attachment 16190 [details] Layout test results
Comment on attachment 16189 [details] Updated patch v2 Basically looks good, but i find myself wondering about SVGTextChunkWalkerBase and SVGTextChunkWalker, as it seems unnecessary to use both virtual methods *and* function pointers.
i'll r+ this if you can justify the exciting combo of function pointer and virtual methods in SVGTextChunkWalkerBase
(In reply to comment #9) > i'll r+ this if you can justify the exciting combo of function pointer and > virtual methods in SVGTextChunkWalkerBase Hey Oliver, yeah nice combination :-) In fact I added SVGTextChunkWalkerBase later on, as it made it possible to remove the template based implementation of "walkTextChunks" in SVGRootInlineBox.h - I didn't like the fact to have crucial methods sitting in header files. Greetings, Niko
Comment on attachment 16190 [details] Layout test results changes look sane, rubber stamping
Landed in r25370.