Mark Rotteveel a écrit :
In other words, your number has 39 digits and is too big to fit in a signed
int128, and therefor is too big for NUMERIC(38)
BTW, as a DECFLOAT(16), that number would be
2039748092580420E+23, so you would lose +/- 23 digits of precision (more due
to rounding)
Mark
oulala, I did not check this
it's my bullshit day.
Many thanks for your time
--
Norbert Saint Georges
http://tetrasys.fi
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
