Path: ...!news.mixmin.net!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: Tim Rentsch <tr.17687@z991.linuxsc.com>
Newsgroups: comp.lang.c
Subject: Re: C23 thoughts and opinions
Date: Sat, 25 May 2024 03:01:07 -0700
Organization: A noiseless patient Spider
Lines: 59
Message-ID: <867cfigqho.fsf@linuxsc.com>
References: <v2l828$18v7f$1@dont-email.me> <20240523150226.00007e7d@yahoo.com> <86msoghwoc.fsf@linuxsc.com> <20240524120544.00000a7d@yahoo.com> <86y17zgvs4.fsf@linuxsc.com> <20240524184623.00004e7f@yahoo.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Injection-Date: Sat, 25 May 2024 12:01:09 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="8137c08e2aeaf8da9fb9d9b9c4e0a9a5";
	logging-data="2981814"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+rUgIQDAkgeTJ3gZwpXeNELJoUHo3rY70="
User-Agent: Gnus/5.11 (Gnus v5.11) Emacs/22.4 (gnu/linux)
Cancel-Lock: sha1:5kbuL+njmhfirVOiUpXKz3LPqpc=
	sha1:8RF5r3cGvHUW5RVI5Tp+PYOXrW4=
Bytes: 3478

Michael S <already5chosen@yahoo.com> writes:

> On Fri, 24 May 2024 06:54:35 -0700
> Tim Rentsch <tr.17687@z991.linuxsc.com> wrote:
>
>> Michael S <already5chosen@yahoo.com> writes:
>>
>>> On Thu, 23 May 2024 17:37:39 -0700
>>> Tim Rentsch <tr.17687@z991.linuxsc.com> wrote:
>>>
>>>> Michael S <already5chosen@yahoo.com> writes:
>>>>
>>>>> [...]  Just want to say that strfrom* family is long overdue, but
>>>>> still appear incomplete.  The guiding principle should be that all
>>>>> format specifiers available in printf() with sole exception of %s
>>>>> should be provided as strfrom* as well.
>>>>
>>>> What's the motivation for having separate functions?  To me this
>>>> looks like creeping featuritis.
>>>
>>> My practical motivation is space-constrained environments, where I
>>> possibly want one or two or three formatters.  sprintf() gives me
>>> all or nothing and all can be too expensive.  Many embedded
>>> environments have big and small variants of sprintf that can be
>>> chosen at link time, but what's in small variant does not
>>> necessarily match a set that I want in my specific project.  And is
>>> not necessarily well documented.
>>
>> Okay, I see now where you're coming from, although I'm not sure that
>> the strfrom*() functions will give you what you want (in terms of
>> memory footprint, etc).  But I get your motivation.
>>
>> Question:  which of the four formats (%A, %E, %F, %G) are ones you
>> expect to use?
>
> Rarely: any of those,  mostly for debugging.
> In productioon code: %e is most likely, but %f could happen.

If you can get by without %g, I recommend writing your own.  The
effort needed isn't trivial but it isn't impossibly large either.
(If you really need %g that's a whole other kettle of fish... and
really old smelly fish at that. :)

> But it's not just a floating point.  "Small" variants of sprintf()
> on 32-bit platforms often unable to handle %lld and %llu.

Here again, just write them.  Easy as falling off a log.


>>  Also I'm curious:  do all of your target platforms
>> use IEEE floating point, or do some use other representations?
>
> Currently, only IEEE.  [...]

My comments above are predicated on being able to count on
floating point being in IEEE format.

Oh, if you want more information about this, please feel free
to email me.