Re: [9fans] rc of Unix

Re: [9fans] rc of Unix

[David Presotto]

Guys, read is not a hard program to write.  Copy the
plan 9 one or write your own.  Should have taken a lot less time
than these messages about sed 1q.

Re: [9fans] rc of Unix

[Kenji Arisawa]

Hello,

I introduced builtin function "read" into Unix rc.

bash$ diff builtins.c builtins.c.orig
30,34d29
< #define RC_READ 1
< #if RC_READ
< static void b_read(char **);
< #endif
<
46,48d40
< #if RC_READ
<       { b_read, "read" },
< #endif
111,136d102
<
< #if RC_READ
< /* read that simulate Plan 9 read -- introduced by Kenar */
< static void b_read(char **av) {
<       char buf[1024];
<       int i = 0;
<       if (*++av != NULL) {
<               set(FALSE);
<               return;
<       }
<       for(;;){
<               if(i == sizeof(buf)){
<                       write(1, buf, i);
<                       i = 0;
<               }
<               if(read(0, &buf[i++], 1) <= 0){
<                       i--;
<                       break;
<               }
<               if(buf[i-1] == '\n')
<                       break;
<       }
<       write(1, buf, i);
<       set(TRUE);
< }
< #endif

where "builtins.c.orig"  is of rc-1.7
I hope that does not introduce a bug ( not  tested enough).

Kenji Arisawa

Re: [9fans] rc of Unix

[rob pike, esq.]

can't resist. here is slightly better code.

/* read that simulate Plan 9 read -- introduced by Kenar */
static void b_read(char **av) {
	char buf[1024];
	if (*++av != NULL) {
		set(FALSE);
		return;
	}
	int i = 0;
	for(;;){
		if(i == sizeof(buf)){
			write(1, buf, i);
			i = 0;
		}
		if(read(0, &buf[i], 1) <= 0)
			break;
		if(buf[i++] == '\n')
			break;
	}
	if(i > 0)
		write(1, buf, i);
	set(TRUE);
}

Re: [9fans] rc of Unix

[boyd, rounin]

> I introduced builtin function "read" into Unix rc.

bad move, human ...

Re: [9fans] rc of Unix

[boyd, rounin]

i used rc on ultrix for 2+ years and i never really ran into a problem.

yes, sed 1q is different to to plan 9's read 'cos sed 1q discards
the whole input stream whereas read just takes the first line
(preserving the the rest of the input stream).

i don't remember how i avoided this problem, but it probably
involved pipe trickery and awk.  it depends on the problem.

Re: [9fans] rc of Unix

[Kenji Arisawa]

Hello,

I said:
> bash$ rc
> ; bash$ rc
> ; read
> read not found
> ;
>
> What should I do for "read" in Unix rc ?
>

Ellis:
 > sed 1q

Thinks, Ellis.
However "sed 1q" behaves somewhat differently.

In the bellow my input sequence are:
aaa
bbb
ccc
ddd


Plan 9 read:
cpu% read; read
aaa
aaa
bbb
bbb
cpu%

Unix "sed 1q":
bash$ rc
; sed 1q; sed 1q
aaa
bbb
aaa
ccc
ddd
ccc
;


Kenji Arisawa

Re: [9fans] rc of Unix

[Bruce Ellis]

sed 1q

----- Original Message -----
From: "Kenji Arisawa" <arisawa@ar.aichi-u.ac.jp>
To: <9fans@cse.psu.edu>
Sent: Saturday, July 05, 2003 12:17 PM
Subject: [9fans] rc of Unix


> Hello,
>
> bash$ rc
> ; bash$ rc
> ; read
> read not found
> ;
>
> What should I do for "read" in Unix rc ?
>
> Kenji Arisawa

[9fans] rc of Unix

[Kenji Arisawa]

Hello,

bash$ rc
; bash$ rc
; read
read not found
;

What should I do for "read" in Unix rc ?

Kenji Arisawa