Hi Paul,
The size of your patch - more than 300 KB - makes me think that perhaps
something is wrong with this part of the GNU standards. Yes, it is formally
incorrect to write foo() when foo takes more than 0 arguments. But it is
a convention that is firmly rooted in Unix tradition.
Anyway, I see many places where removing the parentheses takes away information:
- for alloca, the distinction between the alloca module and the alloca function,
- for iconv, the distinction between the iconv function and the iconv program,
- in lib/sysexits.in.h,
etc.
In other places I prefer the 'quotes' notation. Like here:
the gettext and ngettext macros. This is an alternative to calling
textdomain, and is useful for libraries.
In other places, like doc/functions/errno.texi, there is no need to
remove the parentheses because there the function really takes 0 arguments.
In other places, you left in some ().
So, a question:
Do you know / can you imagine another convention that can be used
to denote a function or macro name as a symbol? Like we use <...>
for URLs or system header files, upper case to denote a variable in
plain text (in place of <var>...</var>), *...* for emphasis, etc.?
Otherwise, I'll rework your patch to lose less information and to please me
better (since much of the code or doc that the patch touches was written by
me).
Bruno
