Camapign for 64-bit T-shirt.

[Malcolm McLean]

I am now taking orders for the campaign for 64 bits T-shirt. The T-shirts,
XXL and in white (one size fits all) will sport the slogan "give me 64".

Wear one to show your support for 64 bit ints.

 

[jacob navia]

I am now taking orders for the campaign for 64 bits T-shirt. The
T-shirts, XXL and in white (one size fits all) will sport the slogan
"give me 64".

Wear one to show your support for 64 bit ints.

Why would anyone want a 64 bit int when a 32 bit is enough for
most purposes? Why not use a long long for 64 bit data?

 

[Keith Thompson]

Why would anyone want a 64 bit int when a 32 bit is enough for
most purposes? Why not use a long long for 64 bit data?

See Malcolm's previous posts on this topic (a Google Groups search for
"compaign", author "Malcolm McLean", newsgroup comp.lang.c will
probably find most of them, though I haven't bothered to check).

Don't expect to change his mind.

 

[Army1987]

I am now taking orders for the campaign for 64 bits T-shirt. The T-shirts,
XXL and in white (one size fits all) will sport the slogan "give me 64".

Wear one to show your support for 64 bit ints.

Why should a 1.50 m tall person weighing 40 kg need a XXL T-shirt?
Why should a variable intended to be a boolean flag, an enum
constant, an error code, a "character or EOF", or similar, have 64
bits? 16 bits should suffice, but if a processor finds it easier
to use 32, that's ok (I'd use short if I needed to pack half a
million integers in as much space as possible). But if it has to
juggle to emulate support for 64 bits for numbers which will never
exceed 30000...
The standard says "natural size". While this is not the same as
requiring int to be int_fast16_t, I guess the intention is
similar.

A more useful campaign would be requiring ULONG_MAX >= SIZE_MAX,
LONG_MIN <= PTRDIFF_MIN, LONG_MAX >= PTRDIFF_MAX, for
compatibility with C89 programs which assume that.

 

[Chris Hills]

I am now taking orders for the campaign for 64 bits T-shirt. The
T-shirts, XXL and in white (one size fits all) will sport the slogan
"give me 64".

Wear one to show your support for 64 bit ints.

Why "support for 64 bit ints" ?

Most of the MCU I use are 8 bit ones.

The vast majority of the worlds MCU are 8 bit I think the 32 bit are
catching up on the 16 bit ones and there are a few 84 and 128 bit ones
out there,.

 

[Robert Gamble]

I am now taking orders for the campaign for 64 bits T-shirt. The T-shirts,
XXL and in white (one size fits all) will sport the slogan "give me 64".

Wear one to show your support for 64 bit ints.

I have 2 questions. First, just out of curiosity, how many orders
have you taken?

Second, is there a place (perhaps on your website, I was not able to
locate anything) where you have laid out a clear and reasoned argument
to support your campaign? I am sincerely trying to understand your
position but I haven't yet seen you provide a convincing argument
(which isn't to say you haven't, I just don't remember seeing one) and
I don't feel like searching for every post you have written on the
subject and trying to piece together the details. There is a white
paper at unix.org (http://www.unix.org/whitepapers/64bit.html) which
describes in some detail the issues and benefits with each of the
different 64-bit data models including the one you propose and
presents a list of reasons why most (UNIX) vendors choose to go the
LP64 route over 10 years ago. Have you produced something similar
that explains why ILP64 is a superior model and perhaps addresses and
counters the points made in the above-mentioned white paper?

Robert Gamble