[Webkit-unassigned] [Bug 67700] REGRESSION(65665): Pattern size being clamped to SVG size can prevent transformed elements from being fully covered by userSpaceOnUse patterns

Thu Sep 15 01:30:23 PDT 2011

https://bugs.webkit.org/show_bug.cgi?id=67700

--- Comment #12 from Nikolas Zimmermann <zimmermann at kde.org>  2011-09-15 01:30:23 PST ---
(In reply to comment #9)
> Really, is this restriction actually *useful*? I can still create a pattern tile that consumes inordinate amounts of memory by just making the <svg> really big (and by "can", I mean my computer is somewhat difficult to use right now because of WebKit thrashing it in the background...)
That's true, I indeed didn't consider that!

>  In any case, this will only be a policy change. Is there something better we can clamp against? Does this restriction even make sense?
We need to clamp against some value, otherwise it's going to trash us - we have testcases that would hang/crash DRT in svg/custom IIRC.

(In reply to comment #11)
> No, I wouldn't expect the spec to, I think this is an implementation detail (trying to prevent people from allocating massive tiles) which was never meant to have a visual impact.
Exactly, it was meant as protection as I disliked hardcoding arbitary values like kMaxSize 1000x1000.
I think the only proper way to handle this is:
- force an upper limit of image buffer sizes
- if requested image size is smaller, compute scale factors that have to be applied before painting to the image context, to make it fit into that size (similar code exists for <filter>)

What do you think?

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.