Re: [9fans] C99

Re: [9fans] C99

[rob pike]

	Is there any plan to update ape/cc to be C99 compliant?

Russ is being a little opaque.  There is no plan to update the
compilers, but there's no plan not to update, either.

-rob

[9fans] C99

[Joel “chesky” Salomon]

According to /sys/src/cmd/cc/c99, among the C99 features implemented in 8c et al is:
	12. Allow declarations anywhere.
Is there a switch to turn that option on?  I recall getting a compile error when I’ve tried that.

--Joel

Re: [9fans] C99

[erik quanstrom]

can you give an example?  this works me.  e.g.

{
	int x;

	x = 1;
	int y;

	y = 1;
	print("%d\n", x+y);
}

or

	for(int x = 1; x < max; x++)
		;

- erik

Re: [9fans] C99

[okamoto]

[-- Attachment #1: Type: text/plain, Size: 285 bytes --]

>http://www.dkuug.dk/JTC1/SC22/WG14/www/docs/n843.ps.gz

I tried this, and found why many Japanese programmers insist "main
must return int".  :-)  The locale.h would also be very nice to someone
else who dont' want to accept unicode.  Anyway, what's the hell of this?

Kenji


[-- Attachment #2: Type: message/rfc822, Size: 1537 bytes --]

From: "Russ Cox" <rsc@plan9.bell-labs.com>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] C99
Date: Tue, 27 Feb 2001 09:25:40 -0500
Message-ID: <20010227142543.C81F2199EE@mail.cse.psu.edu>

http://www.dkuug.dk/JTC1/SC22/WG14/www/docs/n843.ps.gz

Re: [9fans] C99

[forsyth]

[-- Attachment #1: Type: text/plain, Size: 56 bytes --]

what have they done this time, just out of interest?


[-- Attachment #2: Type: message/rfc822, Size: 1702 bytes --]

To: cse.psu.edu!9fans
Subject: [9fans] C99
Date: Tue, 27 Feb 2001 10:43:32 GMT
Message-ID: <3A9B3DFA.6DF08D44@hotmail.com>

Is there any plan to update ape/cc to be C99 compliant?

	david

--
FORTRAN was the language of choice
for the same reason that three-legged races are popular.
	-- Ken Thompson, "Reflections on Trusting Trust"

Re: [9fans] C99

[Russ Cox]

http://www.dkuug.dk/JTC1/SC22/WG14/www/docs/n843.ps.gz

Re: [9fans] C99

[Douglas A. Gwyn]

Russ Cox wrote:
> http://www.dkuug.dk/JTC1/SC22/WG14/www/docs/n843.ps.gz

That's not an accurate copy of the 1999 C standard.
Jack Klein just posted information in comp.lang.c.moderated
about obtaining a correct copy.

There are several sources for a summary of the main changes
from C89 to C99, e.g.
http://web.onetelnet.ch/~twolf/tw/c/c9x_changes.html

The preprocessor extensions are easy enough to implement,
except possibly for the variable-number-of-arguments facility.

Several of the extensions to the language proper are either
already in Plan 9's compiler (compound literals, designated
initializers) or else are straightforward to add.  Variable-length
arrays might be harder.

Most of the library extensions are straightforward to add.
For a start at providing some C99 standard library facilities,
e.g. <stdbool.h> and <stdint.h>, see
http://www.lysator.liu.se/c/q8/index.html
The most difficult one is undoubtedly <tgmath.h>, which
requires compiler support.

My opinion as a long-time implementor and maintainer of C
environments is that *most* of the changes from C89 to C99
can be implemented merely by taking the time to do it,
without having to think very hard, and just a few features
require special care (VLAs, <tgmath.h>).  Until C99 conforming
implementations reach the marketplace and start to be used,
there won't be any demand for the new features.  That means
in probably another year or two it would be useful to have C99
support ready.

Re: [9fans] C99

[Russ Cox]

	Is there any plan to update ape/cc to be C99 compliant?

Whoops.  I misread this, thinking it said "way" not "plan".

A slightly helpful answer: I printed the spec and got _very_
scared away.  In my cynical view, it's not clear how much
good it would do anyway.  The even larger can of worms is the
Single Unix spec.  I doubt the number of exactly conforming
C99/SUSV programs is much higher than the number of exactly
conforming ANSI/POSIX programs currently in existence.

A few weeks ago, I rewrote u9fs with the Single Unix spec
in hand, so that it would just compile everywhere without
any changes whatsoever.  But it was not to be.  As usual,
vendors' implementations vary in subtle ways so that we're
still not at "write once, compile & run anywhere" in any
real sense.

Russ

Re: [9fans] C99

[Russ Cox]

	Is there any plan to update ape/cc to be C99 compliant?

Only the obvious one.

Russ

[9fans] C99

[David Rubin]

Is there any plan to update ape/cc to be C99 compliant?

	david

--
FORTRAN was the language of choice
for the same reason that three-legged races are popular.
	-- Ken Thompson, "Reflections on Trusting Trust"