I was writing some code today that mixed some different sizes of integers, when I realized that the names for the integer types were kind of.. well, weird.

D gets its names for integral types from C.  This isn't a very good idea for 2 reasons:

1, C's integer names tend to be qualitative rather than quantitative; that is, names like "short," "long," and "int" may or may not make sense when developing for processors with different bit sizes.  On a 64 bit processor, for example, "long" doesn't really make sense, as it's the native integer size; it's not long, it's regular.  The only one that makes sense is "byte."

2, D's integer names are subtly different from C's.  In C, "int" is the native integer size, while in D, it is always 32 bits.  In C, long is 32 bits, but it's 64 bits in D.

So I was thinking it would be better if D used quantitative names; that is, the names tell you the size of the integer.  So it would use int8, int16, int32, int64 etc.  This also has the benefit of being expandable - need a 128-bit integer?  Don't use some nonsensical keyword like "cent," use int128.

I know, it's a bit late in the game for something as fundamental as a name change for some of the most basic variable types, but it'd be cool.

Jarrett Billingsley schrieb am Fri, 15 Apr 2005 17:32:08 -0400:
> I was writing some code today that mixed some different sizes of integers, when I realized that the names for the integer types were kind of.. well, weird.
>
> D gets its names for integral types from C.  This isn't a very good idea for 2 reasons:
>
> 1, C's integer names tend to be qualitative rather than quantitative; that is, names like "short," "long," and "int" may or may not make sense when developing for processors with different bit sizes.  On a 64 bit processor, for example, "long" doesn't really make sense, as it's the native integer size; it's not long, it's regular.  The only one that makes sense is "byte."
>
> 2, D's integer names are subtly different from C's.  In C, "int" is the native integer size, while in D, it is always 32 bits.  In C, long is 32 bits, but it's 64 bits in D.
>
> So I was thinking it would be better if D used quantitative names; that is, the names tell you the size of the integer.  So it would use int8, int16, int32, int64 etc.  This also has the benefit of being expandable - need a 128-bit integer?  Don't use some nonsensical keyword like "cent," use int128.

-> dmd/src/phobos/std/stdint.d

"Thomas Kuehne" <thomas-dloop@kuehne.thisisspam.cn> wrote in message news:bvo5j2-407.ln1@lnews.kuehne.cn...
> -> dmd/src/phobos/std/stdint.d

Well, nevermind then!