Reference: #297, #227, #105.

Thanks and thumbs up!

Well this adds a good chunk of overhead. For each single cell, the following pseudo code needs to run in addition to the existing code:

if not locally graylisted
    if globally whitelisted
        return 'gdt'
    if globally blacklisted
        return 'rdt'

And a lot of cell may be visited when going up the matrix inheritance path, all the way up to the all cell.

And this not counting the complication of graylisting cells, like for example user graylist a cell in local scope in order to cancel specific rule in global scope, than later remove said rule from global scope leaving the graylisted cell uselessly explicitly graylisted.

Sigh...

Closing as "won't fix". Sorry, the required changes are scary. On the surface all is simple and easy. Digging under the surface, it's horrible. So we will have to accept the extension can't and won't do everything that comes to mind.

So back again to the same point: scopes are sandboxes, and I can't argue about this anymore. My only suggested alternative from this point forward is to fork the code, I can't implement this without compromising the sanity of the code (and mine I suppose).