Re: AC_DEFINE cleanup

[Stepan Kasal]

Hi,

On Fri, Dec 03, 2004 at 01:07:09PM -0800, Paul Eggert wrote:
> Stepan Kasal <address@hidden> writes:
> 
> > address@hidden:@define] $1 m4_if($#, 2, [$2], $#, 3, [$2], 1)
> > +[#define] $2 m4_if($#, 2, 1, [$3])
> 
> Looks good basically, but why did you get rid of the quadrigraph?
> Presumably it was put in for a reason.

I had no special reason to remove the quadrigraph.
``make check'' didn't complained about this, so if there was a reason,
the testsuite doesn't have it recorded.

I can imagine that the quadrigraph can help in case that some argument
is underquoted.  (Remember, `define' without arguments is not recognized
by GNU m4.)  Such autoconf code would be wrong; do we want to support
it?  If yes, should a test case cover that?

> The proposed ChangeLog entry didn't mention this issue.

So there are two possibilities:
1) to put the quadrigraph back (adding a comment about it, perhaps)
2) to remove the quadrigraph and mention it in the ChengeLog.

Another issue: I have changed the m4_if(...) -- perhaps it should also
be mentioned in the ChangeLog?

I'll wait for your answer and then resubmit the patch.

Thanks,
        Stepan