`SIZE_MAX` is not representable as `double`

[aitap]

Found by the lin-dev-clang-cran check:

freadR.c:835:21: warning: implicit conversion from 'unsigned long' to 'double' changes value from 18446744073709551615 to 18446744073709551616 [-Wimplicit-const-int-float-conversion]

This doesn't break anything right now, but it's a correctness issue.

[venom1204]

Hi @aitap,
Regarding the SIZE_MAX warning in freadR.c:835: since

• 2^64−1 rounds to 2^64 when converted to a double, how would you like this handled?
• Are you looking for an explicit cast (double)SIZE_MAX to silence the warning, or would you prefer using the constant 18446744073709551616.0 (2^64 ) for the comparison instead?

Best regards,

[aitap]

Hi @venom1204! @ben-schwen fixed this in #7767, adding an explicit cast and changing the comparison to >= to make sure that an overflow doesn't happen.