Grisu._show() with 1 digit precision shows 2 digits

[nalimilan]

This behavior may be misleading since I would consider that 2.0e-12 means the 0 is significant, which isn't the case with the second command below.

julia> sprint(Base.Grisu._show, 2.45e-12, Base.Grisu.PRECISION, 2, true)
"2.4e-12"

julia> sprint(Base.Grisu._show, 2.45e-12, Base.Grisu.PRECISION, 1, true)
"2.0e-12"

On the other hand, @printf (which uses a different definition for the precision) returns:

julia> @printf("%.0e", 2.45e-12)
2e-12

I guess an argument in favor of Grisu's behavior is that one gets a standard representation where 2.0 is necessarily a float, i.e. it cannot be mistakenly parsed back as an integer before the parser finds out there's an exponent part. But float("2e-12") works, so that's apparently not a problem.

[jiahao]

Oddly similar to #6608

[nalimilan]

Ah, indeed, this must be a duplicate.