Against PEP 240

Marcin 'Qrczak' Kowalczyk qrczak at knm.org.pl
Thu May 31 19:05:16 CEST 2001


Wed, 30 May 2001 16:26:35 GMT, Terry Reedy <tjreedy at home.com> pisze:

> I wonder why we do not have (widespread, standardized) decimal FPUs
> as well as binary FPUs, which would push to extra overhead to the
> lowest level possible and give a pan-language solution.

Given FPUs available, I don't see an advantage of decimal floating
point over rationals. They are able to represent strictly fewer
numbers. Changing binary to decimal helps for representing 1/5 but not
for 1/3. There are still rounding errors if the number of significant
digits is kept bounded. We pay a performance penalty of software
emulation anyway.

-- 
 __("<  Marcin Kowalczyk * qrczak at knm.org.pl http://qrczak.ids.net.pl/
 \__/
  ^^                      SYGNATURA ZASTĘPCZA
QRCZAK



More information about the Python-list mailing list