• dsimcha (3/3) Nov 28 2008 Should the function signature of std.math.poly be (real x, const real[] ...
• Walter Bright (2/5) Nov 28 2008 Yes, it should be const.
• Denis Koroskin (5/12) Nov 29 2008 crc32 module also needs to be updated. Not only it doesn't use const (bu...

dsimcha <dsimcha yahoo.com> writes:

Should the function signature of std.math.poly be (real x, const real[] A),
not (real x, real[] A)?  The D code at the end of the function clearly doesn't
modify A, but I have no idea what the ASM code is doing.  Is this a bug?

Walter Bright <newshound1 digitalmars.com> writes:

dsimcha wrote:
 Should the function signature of std.math.poly be (real x, const real[] A),
 not (real x, real[] A)?  The D code at the end of the function clearly doesn't
 modify A, but I have no idea what the ASM code is doing.  Is this a bug?

Yes, it should be const.

"Denis Koroskin" <2korden gmail.com> writes:

On Sat, 29 Nov 2008 08:22:34 +0300, Walter Bright  
<newshound1 digitalmars.com> wrote:

 dsimcha wrote:
 Should the function signature of std.math.poly be (real x, const real[]  
 A),
 not (real x, real[] A)?  The D code at the end of the function clearly  
 doesn't
 modify A, but I have no idea what the ASM code is doing.  Is this a bug?

 Yes, it should be const.

crc32 module also needs to be updated. Not only it doesn't use const (but  
should, it allows compile-time crc32 evaluation), but it also uses char[]  
instead of string! :)