Whew - that is quite a read. But I think that this is what I was missing. Thanks Johnathan. So basically, if the compiler is smart enough to ascertain that the result will be unchanged even without the promotions that may be implied by other parts of the standard - then there are no issues (w.r.t. ansi) with optimizing. I guess that this is what the others (who undoubtedly know more about compiler design than me) were frustratedly trying to tell me. The reference was a great help. Thanks again. -----Original Message----- From: Jonathan Wakely [mailto:jwakely.gcc@xxxxxxxxx] Sent: Wednesday, May 28, 2014 2:47 PM To: Regan, Brian (EPC COE) Cc: gcc-help@xxxxxxxxxxx Subject: Re: help understanding behaviour of unsuffixed float constants On 28 May 2014 19:43, Regan, Brian (EPC COE) wrote: > Which ANSI section does that violate? Erm, which does the single-precision floating point comparison violate? Please read 5.1.2.3 carefully.
