From mboxrd@z Thu Jan  1 00:00:00 1970
From: erik quanstrom <quanstro@quanstro.net>
Date: Thu, 13 May 2010 09:24:47 -0400
To: 9fans@9fans.net
Message-ID: <158d5dc0571fa04cab44a99b3fdf5921@kw.quanstro.net>
In-Reply-To: <op.vcmv3401hipq0d@santucco.avp.ru>
References: <AANLkTinmbNhBI_V3bvIctfCXdNwzhYNjtoDaA7Mcc3D7@mail.gmail.com>
	<op.vck9apb3hipq0d@santucco.avp.ru>
	<AANLkTin7F2mT7yhs4VQATcvoIsKMDjsYwDzQX2sN6zyV@mail.gmail.com>
	<op.vcmv3401hipq0d@santucco.avp.ru>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Subject: Re: [9fans] hoc output format
Topicbox-Message-UUID: 202ba602-ead6-11e9-9d60-3106f5b1d025

On Thu May 13 03:51:56 EDT 2010, santucco@gmail.com wrote:
> Yes, I'm agree, but with one exception - awk(1) separates a data from a
> code, hoc(1) doesn't do it. So hoc(1) can be used for plain calculation
> tasks, not for  processing input files with a data.

both awk and hoc accept standard input

	echo 1 + 2 | hoc	# (sic, see below)
	echo 1 2 | hoc -e '{while(read(x) != 0)y += x' ^ $nl ^ ' print y, "\n"}' -
	echo 1 2 | awk '{for(i = 1; i <= NF; i++) x+= $i}END{print x}'

or execute directly from the command line

	hoc -e '1 + 2'
	awk 'BEGIN{print 1 + 2; exit}'

note that the wierd extra {} are required for hoc.  hoc read
requests are satisfied from the same file descriptor as the data.
so the input needs to be exhausted before read runs.

also, the trailing newline on the input is required, since Bgetd()
won't accept '1<EOF>'.  this seems like a bug induced by the
fact that charstod has one argument too few.  if it also returned
the number of characters consumed, Bgetd could read

int
Bgetd(Biobufhdr *bp, double *dp)
{
	double d;
	struct bgetd b;

	b.b = bp;
	b.eof = 0;
>	b.nchr = 0;
>	d = charstod(Bgetdf, &b, &n);
>	if(b.eof || b.nchr != n)
		return -1;
	Bungetc(bp);
	*dp = d;
	return 1;
}

- erik