Re: 2.62 AT_SETUP limitations

[Akim Demaille]

Le 22 avr. 08 à 14:45, Eric Blake a écrit :
> Really, the problem is
> that there is no way to determine whether $1 is safe to expand,  
> short of
> expanding $1; thus, if it was not safe, you've already screwed up the
> state of the m4 parsing.

I always felt m4 lacks an "eval" macro.  Not the one for arithmetics,
but one which would run m4 on its argument, and return the expansion
quoted.  This is what tried to do m4_quote/m4_dquote, but it's fragile.

There are really two sort of macros.  Those which really work on the
text, and you do not want to evaluate the result.  This is where
regexp, patsubst etc. live.  And there are the regular just-expanding
macros.

For the latter, a usual single-level of quotation is fine.  For the
former, we need two.  And we have no safe means to go from one model
to the other, between text-hacking macros and text-expanding macros.

Something like a new primitive "eval and return quoted" would fill
the gap.

Well, at least that's what I believed years ago :)