b122db670b
...to avoid UB when fVal was cast to nRet *before* checking that fVal actually
fit into the bounds, and to avoid Clang 10
-Werror,-Wimplicit-int-float-conversion when comparing fVal against
SAL_MAX_INT64 ("implicit conversion from 'const sal_Int64' (aka 'const long') to
'double' changes value from 9223372036854775807 to 9223372036854775808")
Change-Id: I9e2f6c97309609d9ec2455d4ecf9c341d85c1680
Reviewed-on: https://gerrit.libreoffice.org/77430
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>