Date:	Fri, 29 Jun 2001 12:50:11 +0300
From:	"Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il>
Sender:	halo1 AT zahav DOT net DOT il
To:	"Mark E." <snowball3 AT bigfoot DOT com>
Message-Id:	<8361-Fri29Jun2001125011+0300-eliz@is.elta.co.il>
X-Mailer:	Emacs 20.6 (via feedmail 8.3.emacs20_6 I) and Blat ver 1.8.9
CC:	djgpp-workers AT delorie DOT com
In-reply-to:	<3B3BC9AE.12.3CD099@localhost> (snowball3@bigfoot.com)
Subject:	Re: v2loadimage and proxy
References:	<3B39E3C5 DOT 6290 DOT 765DBB AT localhost> (snowball3 AT bigfoot DOT com) <3B3BC9AE DOT 12 DOT 3CD099 AT localhost>
Reply-To:	djgpp-workers AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

> From: "Mark E." <snowball3 AT bigfoot DOT com>
> Date: Fri, 29 Jun 2001 00:19:58 -0400
> 
> I've incorporated your info into this revision:

Thanks.  I have only two comments:

> +   while (1)
> +   {
> +     if (*ptr == '\r')
> +       break;

This means command lines cannot have embedded CR characters, which
night be a subtle misfeature.  I think get_arg should accept another
argument, the actual length of the command line, and only stop at the
CR if it is at the expected end of the command.

But that begets a new problem: the first byte of cmdline cannot pass
more info than about 256 bytes.  So if we want to support embedded
CRs, we need to extend the format of cmdline passed to v2loadimage in
some way.  For example, we could let command lines have a 0xFF in the
first byte, and then they need to be null-terminated strings, in which
case the length will be taken from strlen.

I know that this is a bit of a nuisance to code, but I don't like to
have limitations like no embedded CRs, especially if we now extend
v2loadimage to support very long commands.

However, if others think this is too much of a hassle for supporting
CRs, I won't insist (provided it's documented).

> +   if (proxy_made)
> +   {
> +     /* When the command line is long enough to justify a proxy variable,
> +        set the command tail accordingly.  */
> +     memcpy(proxy_cmdline, cmdline, 128);
> +     proxy_cmdline[0] = 126;
> +     proxy_cmdline[127] = '\r';
> +     cmdline = (const char *)proxy_cmdline;

This last line seems to be a mistake, because you later have:

> +   /* Setup the buffer with proxy arguments.  */
> +   if (proxy_made)
> +     make_proxy_buffer(program, cmdline, proxy_argc, __tb, __tb_size);

And you want the original command line to be passed to
make_proxy_buffer, I think.  (Actually, cmdline+1.)

Last, but not least, the docs should be augmented to reflect these
changes.

Thanks again for working on this!

- Raw text -