isAntialiased() accessing non-sibling pixels/negative offsets on image boundaries

[radomik]

In function isAntialiased() there are read pixels which are direct siblings of sourcePix:

offset = ((verticalPos+j)*width + (horizontalPos+i)) * 4;
if(!getPixelInfo(targetPix , data, offset, cacheSet)){
    continue;
}

Pixel sourcePix has coordinates (horizontalPos,verticalPos) and i, j can be either -1 or 1. Using that values there is computed offset used in getPixelInfo():

function getPixelInfo(dst, data, offset, cacheSet) {
    if (data.length > offset) {
        dst.r = data[offset];
        dst.g = data[offset + 1];
        dst.b = data[offset + 2];
        dst.a = data[offset + 3];
        return true;
    }
    return false;
}

Function getPixelInfo() access pixel data stored row-wise.

When (for example) isAntialiased() takes sourcePix positioned on the end of a row then, with i=1 a sibling pixel is taken from the beginning of following image row which is not correct.
Similar problem will occur when sourcePix is placed on the row beginning, then siblings are taken from the end of previous row.
There may be also needed to add a check for negative offsets i.e. when sourcePix is positioned on the first image row (now there is only a check for too big offsets - in getPixelInfo()).

[jamescryer]

I think you're right, and it is solvable by breaking out of the loop in isAntialiased(). Unfortunately the same check seems to affect performance, I had mixed results on testing. For that reason I'm not going to fix this issue, unless it's causing any problems?