On Sun, Apr 07, 2002 at 09:16:19AM -0700, Adam J. Richter wrote: > >there's no good reason for autoconf to start poaching on symbols that > >don't begin with AC_. File it as a bug in autoconf. > > Recent versions of autoconf appear to split a lot of the autconf > functionality between autoconf proper and "m4 sugar." I think the > idea is that the m4sugar code could be useful for other things, so > that why I think they want a second prefix. Personally, I think
it would be nice if they documented it (there are 8 lines in the doc directory which deal with that suffix, none are in an overview - the changelog lists numerous instances of changes, but the changelog is worthless because it seldom mentions why a change was made, or its impact - less informative than a phonebook). > that second prefix should be something less likely to collide with > existing variable (like M4S_, say), and that autoconf should just > generate warning about this since it does not know what variables > people have used up to this point. > > In general, I think a number of recent changes in autoconf > have broken backward compatability unnecessarily. (true ;-) > Nevertheless, autoconf-2.53 is out there. You will probably > have to field the same trouble report repeatedly if you do not work > around the problem. The workaround is not only very simple; it's also I don't have an AS_ prefix, but the warning is well-taken since there's no guarantee based on past experience that autoconf wont' grab off more of the namespace). Past experience shows that anytime they want to change the design, they declare the older version to have bugs (hence disregard) or "undocumented". -- Thomas E. Dickey <[EMAIL PROTECTED]> http://invisible-island.net ftp://invisible-island.net
