[Webkit-unassigned] [Bug 40793] [GTK] The size of the shadow image uses the standard deviation size instead of the blur radius

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=40793





--- Comment #6 from Alejandro G. Castro <alex at igalia.com>  2010-07-02 00:35:01 PST ---
(In reply to comment #5)
>
> [...]
>
> I'm quite sure, that this need updates of pixel tests. Can you run layout tests with a tolerance of zero please?

Yep, the visual result is different, I was wondering if other platforms with pixel tests for this filter could have problems because I could not test it in webkitgtk+. I'll review what we need.

> I'm also not sure about kernelPosition. The spec for feGaussianBlur changed recently and a blurring radius of 0 is allowed now. If a stdDeviation for any direction is zero, the image is just not blurred for this direction. At the moment we break and don't draw the blurring at all.

I understand, we can change also this condition that controls the standard deviation parameter:

    if (m_x == 0 || m_y == 0)
        return;

with

    if (m_x == 0 && m_y == 0)
        return;

And add ifs to the calculations to avoid calculating sdx or sdy and avoid the blox blur in that direction.

Probably the name of the parameters/variables are still confusing, I could change m_x and m_y with mStdx and mStdy, and sdx/sdy with kernelSizex/kernelSizey.

> Now that you try to fix the SVG filter as well, and I'm realy glad about this :-), can you take about this fact as well? It might also be possible to move the feGaussianBlur fixes to another patch,if you think this is not the right place.

Yep, actually it was my my initial intention to do it separately, but finally checking the visual result of reviewing the kernel_size I've checked the position of the shadow was not correct without all the modifications, and I was not sure if it was due to the change I did, my fault, I'll check it and try to separate both patches adding the kernelPosition. In case we have pixel tests we will have to do two modifications also.

> Great patch! Please fix the style problems and think about the last comment about kernelPosition. r- for now, also for breaking Mac.

I'll do, thanks very much for the review Dirk :).

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