We don't rely on the mask for insulation, we rely on a 7 mil clearance
copper-to-copper, but that leaves a smaller mask width between them.
For dog bones we do similar, but pad-to-via is the same potential, and
those vias are encroached with mask anyway.
not feeling any shame, but thanks for writing back.
(This report also flags "copper slivers" which I've never heard of, but to
them it is the thin copper plane that floods between vias!)

On Tue, Feb 19, 2019 at 10:45 AM Stadem, Richard D <[log in to unmask]>
wrote:

> For soldered BGA dog-bone pads, that is a major issue, common within the
> industry. If the soldermask strip between the via and the BGA pad is too
> small, there is not enough adhesion and the molten solder runs right under
> the soldermask strip down into the via hole. Too thin of a strip of
> soldermask allows solder thieving from the solder ball. But for unsoldered
> vias not connected to a soldered pad on the opposite side, I cannot imagine
> what issues there would be. Soldermask is not considered an electrical
> insulator, so if a designer was depending on a thin sliver of mask to
> provide insulation between two vias of different potential, shame on them.
>
> -----Original Message-----
> From: TechNet [mailto:[log in to unmask]] On Behalf Of Jack Olson
> Sent: Tuesday, February 19, 2019 9:58 AM
> To: [log in to unmask]
> Subject: [TN] soldermask silvers
>
> In my experience, what a PCB fabricator would consider a sliver would be
> the strip of soldermask material between RECTANGULAR pads being less than a
> minimum width.
> We got a "DFM report" for a board where minimum distance between ROUND vias
> is being reported as slivers.Has anyone ever heard of problems with minimum
> mask material between round, unsoldered vias?
> I can't imagine any real-world problems that could cause, but I have an
> open mind...
>