On 2013-09-19 23:39 +0200, su_v wrote:
r12525: breaks FER position and size: While trying to figure out the current relationship between FER, visual bounding box (and stroke with), I came across another regression in the filter editor: Changing coordinates and/or dimensions in the filter editor does not live-update the filter output area (it only affects the position and size of the visual bbox frame). This seems to have started with the changes in r12362 (in tests with archived builds it works as expected with r12359, in r12364 and later revisions one has to force a refresh or reload the file to see the FER updated), and regressed one more step in r12525 (now AFAICT the FER neither is correctly updated when forcing a refresh nor when reloading the file).
AFAICT still broken with r12544.
Maybe one of the SVG-filter-effects experts (wrt the spec and inkscape's implementation) can test and comment?
Test case uploaded here: https://www.dropbox.com/sh/texl8xxvw77pkjo/LY_NVrWLGp?lst#/
Files: FER-regression-4-batik17.png : reference rendering Squiggle (Batik 1.7)
FER-regression-4-chromium.png : screenshot chromium 31.0.1638.0 (224352)
FER-regression-4-r12359.png : export inkscape r12359 (same as batik & chromium)
FER-regression-4-r12364.png : export inkscape r12364 (visual instead of geometric bounds for FER?)
FER-regression-4-r12524.png : export inkscape r12524 (same as with r12364)
FER-regression-4-r12550.png : export inkscape r12525 (what happened?)
FER-regression-4.svg : sample SVG file