Path: csiph.com!usenet.pasdenom.info!aioe.org!eternal-september.org!feeder.eternal-september.org!mx02.eternal-september.org!.POSTED!not-for-mail
From: Keith Thompson <kst-u@mib.org>
Newsgroups: comp.lang.c
Subject: Re: Spurious diagnostic messages
Date: Sun, 02 Aug 2015 15:31:21 -0700
Organization: None to speak of
Lines: 40
Message-ID: <lnwpxdiapy.fsf@kst-u.example.com>
References: <874mn5ty3y.fsf@kst-u.example.com> <mjna5l$duf$1@dont-email.me> <ejevladml0ep58beoapn0sn3fog7pt4nnv@4ax.com> <87egm8qis7.fsf@kst-u.example.com> <kfnvbfi6nnp.fsf@x-alumni2.alumni.caltech.edu> <mjus0p$4eb$1@dont-email.me> <kfnfv4w53r5.fsf@x-alumni2.alumni.caltech.edu> <molbhc$jss$1@dont-email.me> <kfnfv4bfw05.fsf@x-alumni2.alumni.caltech.edu> <mp2bgi$d84$1@dont-email.me> <kfnbneraq9d.fsf@x-alumni2.alumni.caltech.edu> <d24o1iF55b9U1@mid.individual.net> <51drra5l8kp2okl6m99fpfuppmo5n1g16d@4ax.com> <mpl49o$k76$1@dont-email.me> <mpl5hk$ocu$1@dont-email.me>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Injection-Info: mx02.eternal-september.org; posting-host="945944de09706c9b4e29b53c9d2efdc2"; logging-data="23443"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX185HiaKlenI3xiY0IDzqqo4"
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux)
Cancel-Lock: sha1:zh5xSsXjZZCDyoGDoZfMFrHtjt4= sha1:BltNusi7CPtFtm09dJZqhj3MIn8=
Xref: csiph.com comp.lang.c:66752

Bartc <bc@freeuk.com> writes:
[...]
> What about binary FP constants?
>
> (I tried this in my non-C language:
>
>    print 0B11.101
>
> output when told to translate to C:
>
>    printf("%f",3.625);
>
> Which I was surprised at as I can't remember implementing it. However it 
> needs some attention as 0B11.101e100 doesn't work. It's also not that 
> clear whether the exponent should be a power of 10, or of 2.
>
> Binary FP output is not available as that uses C's sprintf, which 
> doesn't support it.)

C already has hexadecimal floating-point constants, using 'P' rather
than 'E' to mark the exponent (since 'E' is a valid hexadecimal digit).
The exponent represents a power of 2.

*If* binary FP constants were to be added to C, they should be just like
hexadecimal FP constants but with a "0b" prefix rather than "0x".  Thus
0b1.1p3 == 12.0.

I don't suggest that binary FP constants (or I/O) would be worthwhile.
Non-decimal floating-point is not as human-readable as non-decimal
integers, and hexadecimal already serves the purpose of capturing the
exact value (for systems where FLT_RADIX is a power of 2).

(Incidentally, Ada supports numeric constants and I/O using all bases
from 2 to 16.)

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
Working, but not speaking, for JetHead Development, Inc.
"We must do something.  This is something.  Therefore, we must do this."
    -- Antony Jay and Jonathan Lynn, "Yes Minister"