[cfp-interest 3663] Re: [SC22WG14.34140] CFP teleconference - Wednesday, October 15

[Mike Cowlishaw]

It is because floating-point arithmetic is extremely complex and mathematically subtle.  The 'base standard' for it is IEEE 754 (first published in 1985, 40 years ago) and the details are still being refined and improved today (the third, "2029", revision is in progress).   C floating point support was not ideal, initially, and has been hugely improved over the years by the efforts of the CFP working group -- who have had quite a task maintaining backwards-compatibility, while adding new features from IEEE 754 such as decimal floating point (on which our financial systems are totally dependent nowadays).
 
Mike Cowlishaw
mfc at speleotrove.com
https://speleotrove.com/mfc/
 
 


  _____  

From: cfp-interest [mailto:cfp-interest-bounces at oakapple.net] On Behalf Of Marcus Johnson
Sent: 12 October 2025 05:09
To: Robert Seacord
Cc: CFP; SC22 WG14
Subject: [cfp-interest 3650] Re: [SC22WG14.34140] CFP teleconference - Wednesday, October 15


I'm not angry or trying to say it's not worthy or anything, I'm truly just trying to understand why I see so many meetings for it.


On Oct 11, 2025, at 8:53 PM, Robert Seacord <rcseacord at gmail.com> wrote:



 
I think CFP is doing a good job, and it's not costing anyone anything, so what's the beef? 

rCs

On Sat, Oct 11, 2025, 8:13 PM Marcus Johnson <marcusljohnson1991 at gmail.com> wrote:


The latter, why does the C language need so much effort to support floating point types?

Does C++ require this much effort for the same support, or does it get it for free?

Seems like something compiler implementers would need to support, not the language it's self.


On Oct 11, 2025, at 12:48 PM, Jim Thomas <jaswthomas at sbcglobal.net> wrote:



 
The C Abstract says, "Its purpose is to promote portability, reliability, maintainability, and efficient execution of C language programs on a variety of computing systems.” Are you asking why that purpose needs to cover programs that use floating point? Or, are you asking why fulfilling that purpose for floating point needs to be so complicated? Or something else?

- Jim Thomas



On Oct 9, 2025, at 9:02 AM, Marcus Johnson <marcusljohnson1991 at gmail.com> wrote:

Sorry for interrupting, but exactly does C need to devote so much energy to floating point types?

Like, there's the sign bit, the exponent, an implicit bit, and the mantissa.

Why so much effort?


On Oct 9, 2025, at 10:59 AM, Jim Thomas <jaswthomas at sbcglobal.net> wrote:



 
The next teleconference of the C Floating-Point Study Group will be 

Wednesday, October 15, 2025
8 AM PDT / 3 PM UTC

A draft agenda for this teleconference, including directions for joining the Zoom meeting, and minutes from previous teleconferences are posted on the CFP wiki <https://wiki.edg.com/bin/view/CFP/WebHome> (contact me if you need login info).

- Jim Thomas  



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20251016/5966ab58/attachment.htm>