[9front] pcc macro issue

[9front] pcc macro issue

[Xiao-Yong Jin]

The following is wrong.

cpu% cat test.c
#define N
#define F(x)
#define M F(N)
M
cpu% pcc -E test.c
cpp: test.c:4 Disagreement in number of macro arguments
#line 1 "/tmp/test.c"



 F()
pcc: cpp: cpp 25337: errors

Re: [9front] pcc macro issue

[Kurt H Maier]

On Wed, Jan 20, 2021 at 11:59:31PM -0600, Xiao-Yong Jin wrote:
> The following is wrong.

Only in C99. In C89 this is undefined.

khm

Re: [9front] pcc macro issue

[Jonas Amoson]

This is a distilled example. Do you have a real case where
you encountered a problem? I am no expert in C macros,
but it could be interesting to try to see.

On Thu, 21 Jan 2021 at 08:25, Xiao-Yong Jin <meta.jxy@gmail.com> wrote:
>
> The following is wrong.
>
> cpu% cat test.c
> #define N
> #define F(x)
> #define M F(N)
> M
> cpu% pcc -E test.c
> cpp: test.c:4 Disagreement in number of macro arguments
> #line 1 "/tmp/test.c"
>
>
>
>  F()
> pcc: cpp: cpp 25337: errors
>

Re: [9front] pcc macro issue

[Xiao-Yong Jin]

https://github.com/gambit/gambit/blob/ca8b488447f78d4165f3f3064875c57424b42a3e/include/gambit.h.in#L6338-L6380

> On Jan 21, 2021, at 4:48 AM, Jonas Amoson <jonas.amoson@gmail.com> wrote:
> 
> This is a distilled example. Do you have a real case where
> you encountered a problem? I am no expert in C macros,
> but it could be interesting to try to see.
> 
> On Thu, 21 Jan 2021 at 08:25, Xiao-Yong Jin <meta.jxy@gmail.com> wrote:
>> 
>> The following is wrong.
>> 
>> cpu% cat test.c
>> #define N
>> #define F(x)
>> #define M F(N)
>> M
>> cpu% pcc -E test.c
>> cpp: test.c:4 Disagreement in number of macro arguments
>> #line 1 "/tmp/test.c"
>> 
>> 
>> 
>> F()
>> pcc: cpp: cpp 25337: errors
>> 

Re: [9front] pcc macro issue

[ori]

Quoth Xiao-Yong Jin <meta.jxy@gmail.com>:
> The following is wrong.
> 
> cpu% cat test.c
> #define N
> #define F(x)
> #define M F(N)
> M
> cpu% pcc -E test.c
> cpp: test.c:4 Disagreement in number of macro arguments
> #line 1 "/tmp/test.c"
> 
> 
> 
>  F()
> pcc: cpp: cpp 25337: errors
> 

Yep, I fixed a lot of problems in cpp, but this
is one that I didn't get to. Another one is
space handling, which we currently ignore.

	#define MAC(x) y

	MAC(x)	// should expand to y
	MAC (x)	// should remain untouched

ANSI C macro expansion rules are subtle and
arcane. I've got a few tests in my underwhelming
regression suite, which may help if you want
to take a stab at this:

http://shithub.us/git/ori/regress/HEAD/cmd/cpp/f.html