On Mon, Mar 24, 2014 at 01:32:13PM +0100, Bernd Paysan wrote: > > 2) Just like for checking for registers and long longs, try whether > > the engine works without the workaround mentioned above, and only > > employ the workaround if it doesn't. > > The workaround is for historical GCCs; we might also cut support for those.
Even if we wanted to cut support for old gcc versions (which we don't, see below), this puts more faith in the gcc maintainers than they deserve. This problem might come up in the next version of gcc, just like it has come up before. > Who's stlill using GCC 3.x? If you want to use a historical GCC, please use > GCC 2.95.2 (though this one has no -fwrapv switch, and also causes some > troubles ;-). I am still using gcc 3.x, on some platforms that I use for portability testing of Gforth. And gcc 2.95 is not always available and is buggy on some platforms. In any case, we gain very little from removing that support. And since one of our goals is portability (and that includes portability to systems with old gccs), we should keep it even if we knew that gcc will never regress into that problem again. - anton
