From mboxrd@z Thu Jan  1 00:00:00 1970
From: erik quanstrom <quanstro@quanstro.net>
Date: Sat,  7 Sep 2013 09:52:13 -0400
To: 9fans@9fans.net
Message-ID: <3caa188579ff2a78ca7f01d49ae68690@kw.quanstro.net>
In-Reply-To: <522AE229.5070603@gmx.de>
References: <12773354.WkoI8jKRLR@coil>
	<522AE229.5070603@gmx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Subject: Re: [9fans] p9p mk(1) syntax
Topicbox-Message-UUID: 7bbf74d8-ead8-11e9-9d60-3106f5b1d025

On Sat Sep  7 04:23:05 EDT 2013, f.psiorz@gmx.de wrote:
> I don't think that the behavior is really unexpected ... the `{...} part
> is interpreted by mk, but not the text between the braces. You probably
> need to quote it if it contains braces itself.
>

after looking at the code, it appears to me that this is an oversight,
that does not effect bourne-shell compatabiles.  we simply need to
count the `{s and subtract the }s, and when the count reachs 0
we're done.

(it appears that ' quoting is incorrect for bourne shell compatables,
and the interpretation of \ is incorrect for rc.)

- erik

; diffy -c lex.c
/n/dump/2013/0907/sys/src/cmd/mk/lex.c:67,73 - lex.c:67,73
  static int
  bquote(Biobuf *bp, Bufblock *buf)
  {
- 	int c, line, term;
+ 	int c, line, term, depth;
  	int start;

  	line = mkinline;
/n/dump/2013/0907/sys/src/cmd/mk/lex.c:79,88 - lex.c:79,91
  			;
  	} else
  		term = '`';		/* sh style */
+ 	depth = 1;

  	start = buf->current-buf->start;
  	for(;c > 0; c = nextrune(bp, 0)){
- 		if(c == term){
+ 		if(term == '}' && c == '`')
+ 			depth++;
+ 		if(c == term && --depth == 0){
  			insert(buf, '\n');
  			insert(buf,0);
  			buf->current = buf->start+start;