>> I think we should have this in core, as this definitely is a bug. > > hard to say - anything about CHAR(N) is strange,
On a quick scan of the standard, it looks like our current behavior is non-conforming.
> and this change can break existing applications :(
That is true, but since the only point of supporting CHAR(n) is to satisfy requirements of the standard, it might be something we should do, if technically feasible.
It is true, but nobody reported bug yet, so who know, how this feature is used.
Probably it needs a deeper analyze of difference between Pg CHAR(n) and standard CHAR(n)