Re: doco m4 notes

[Kevin Ryde]

Paul Eggert <address@hidden> writes:
>
> It's a nice set of notes, but wouldn't it be better fodder for the GNU
> m4 manual?  It's not really relevant to Autoconf per se.

I thought of autoconf mainly since it's notes about treading that fine
line of portability and bug avoidance.  

> One thought is that you might want also to mention what POSIX
> 1003.1-2001 (with corrigenda) requires for an m4 behavior, if it
> requires one.  See
> <http://www.opengroup.org/onlinepubs/007904975/utilities/m4.html>.

Ah, yep.  I wasn't aware that was available online, I thought you had
to pay big bucks to have a copy.  A nice link for "Systemology" in the
autoconf manual perhaps.

> Kevin Ryde <address@hidden> writes:
>
>> dnl  m4wrap() sequence - in BSD m4, m4wrap() replaces any previous m4wrap()
>> dnl       string, in SysV m4 it appends to it, and in GNU m4 it prepends.
>> dnl       See m4wrap_prepend() below which brings uniformity to this.
>
> POSIX requires the SysV behavior.  Another bug report for GNU m4, perhaps?

Hope nobody has depended on the way gnu m4 works.  Maybe it'd have to
be confined to POSIXLY_CORRECT mode or something.

> Here by "SunOS" I assume you mean "SunOS 4.x".

Yes.  If I work up the notes I might call it V7 m4 or something like
that.  The V7 man page (link in the autoconf manual), lacks those
features.  Doesn't make that m4 completely unusable, but too much like
hard work for what I wanted.

> also, that xpg4 m4 uses
> C-style precedence (but the precedence for ccs m4 is not defined).

I hadn't tried to pay much attention to precedence :-).