Re: status.m4 off-by-one

[Ralf Wildenhues]

Hi Paul,

* Paul Eggert wrote on Tue, Nov 08, 2005 at 10:34:56PM CET:
> I don't quite understand the bug or the patch but you might as well
> install it if it fixes your bug.  Thanks.

This snippet:

 m4_if(_AC_SED_CMD_LIMIT,
    m4_if(_AC_Var, address@hidden@], m4_if(_AC_SED_CMD_NUM, 2, 2, 
_AC_SED_CMD_LIMIT), _AC_SED_CMD_NUM),
 [blurb])

is supposed to encode this:
  expand to blurb, iff (
     ( (_AC_Var equals @END@) and (_AC_SED_CMD_NUM is not equal to 2) )
    or
     (_AC_SED_CMD_LIMIT equals _AC_SED_CMD_NUM)
   ) 

which in turn means, more informally:
  if we've filled one sed script, or we're both at the end of the loop
  over all variables and the current sed script doesn't happen to be
  empty (which means non-existant, due to other logic),
  then output blurb which is the trailer for the script plus its pre-
  postprocessing.

The hacky part is the encoding of this logic, with mentioning blurb only
once; my version requires _AC_SED_CMD_LIMIT to be != 2 for this to work;
this however is needed anyway, so it's not a real limitation.

It surely could be done differently, or, FWIW, by defining a macro
containing blurb, so we can use it more than once easily.  However,
blurb is actually pretty complicated stuff itself, so.

I've applied the patch.

Cheers,
Ralf