On Thu, Nov 08, 2007 at 09:19:25PM +0100, Enrico Forestieri wrote: > On Thu, Nov 08, 2007 at 09:04:37PM +0100, Andre Poenitz wrote: > > On Thu, Nov 08, 2007 at 07:53:13PM +0100, Enrico Forestieri wrote: > > > On Thu, Nov 08, 2007 at 06:13:15PM +0100, Andre Poenitz wrote: > > > > On Thu, Nov 08, 2007 at 01:55:44PM +0100, Enrico Forestieri wrote: > > > > > At least, I would suggest changing > > > > > > > > > > #ifdef _MSC_VER > > > > > > > > > > into > > > > > > > > > > #if ! defined(__GNUC__) > > > > > > > > > > as I think that forward declaring string is a gcc-ism. > > > > > > > > On what foundations did you base your thinking? > > > > > > That should have been read in its context. I agree that I should > > > have said "as I think that forward declaring string in the iosfwd > > > header is a gcc-ism", but I didn't think it would have been so > > > difficult to understand. > > > > I do not want <iosfwd> in strfwd.h. > > Why?
Because it voids 30% of the effect of having strfwd.h at all and it is not needed here and therefore probably on ~100% of the Linux systems out there. Andre'
