lua-users home
lua-l archive

[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index] [Thread Index]

On Fri, Sep 09, 2016 at 05:03:17PM -0300, Roberto Ierusalimschy wrote:
> > This is one reason why (IIRC) the C working group recently rejected making
> > IEEE floating point mandatory. And it's also why they're still reticient to
> > more strictly define integer representations and arithmetic behavior--you
> > never know what advancements might come around the corner.
> The C99 standard does not leave much room for new integer
> representations, as it explicitly lists the allowed representations:
> sign and magnitude, two's complement, and one's complement
> (ISO/IEC 9899:1999 (E), Integer types).

I had in mind padding bits. I was thinking about a comment I stumbled across
several months ago related to signed integers. Now that I thought about it
more, it might have concerned a proposal to make signed overflow
implementation defined. That's not a representation issue. Or maybe it is if
it involves the state of padding bits and possible trap representations.

In my recollection the comment was made by a C or maybe C++ committee
member, but I can't find any such comment Googling the committees' websites
and documents for topics related to signed integers. And definitely no
mention of unums. Maybe it's just my imagination.