Re: bison-1.29c 'configure' problems on Solaris 8.

[Akim Demaille]

>>>>> "Paul" == Paul Eggert <address@hidden> writes:

Paul> I still see many problems with that.  E.g., it doesn't propagate
Paul> the exit status, 

What are you referring to?  If configure.lineno exits somewhere, how
could configure survive?

Paul> it doesn't handle multiple instances of LINENO on the same line,
Paul> it mishandles $LINENOUGH (:-), etc.

:)  Yeah, I know.  But that's exactly the code that was used in
autoconf-2.13, which demonstrates this is enough.  If we can go
further portably, I'm all for it.

Paul> We might get something like that to work, but how about using a
Paul> modern shell instead, if available?  Nowadays, almost everybody
Paul> has a modern shell somewhere -- the problem is that it's not
Paul> called "/bin/sh".  So we can arrange for configure to reexecute
Paul> itself with a POSIX-compliant shell if it discovers that
Paul> "/bin/sh" does not conform to POSIX.  This would help in several
Paul> areas, not just $LINENO.

I *agree*, that's also my plan, but I think this time we are going too
fast.  First let's find a portable LINENO, *then* move to another
shell.

I definitely agree with you on that point.

But, I would like to see that handled in M4sh, *not* Autoconf.  Also,
I would like to provide means to specify what are the requirements
you'd like to see fulfilled by the new shell (e.g., functions).


Paul> The only LINENO-related downside I see for this is that people
Paul> that have only ancient shells will see line number 0 in their
Paul> output.  This is unfortunate, but it's not a fatal objection, as
Paul> the line numbers are not necessary for correct operation.  And
Paul> we can encourage those folks to port bash at their earliest
Paul> convenience.

Agreed, I have been tempted by that thinking several times, but I
stepped back thinking that machine where lineno is not supported are
probably machine where we will have problems, hence, where line nums
are the most useful.

I prefer that we _try_ to change these LINENO, even if it's only one
occurrence per line, than having "" (not 0) everywhere.


As a first step towards correction of the current Autoconf, I apply
the patch I suggested.