Need to learn C, best books?

Peter Corlett abuse at cabal.org.uk
Fri Oct 26 14:26:43 BST 2007


On Fri, Oct 26, 2007 at 11:50:11AM +0100, Dirk Koopman wrote:
[...]
> What stopped this nonsense was the realisation that, actually, processing
> text was at least as, then becoming more, important than number crunching.
> So newer processor designs became directly character addressable -
> although there was still some debate about what a character was in terms
> of storage.

But only for general-purpose processors. Specialist processors like the
TMS320 have their basic unit as the machine word. sizeof(char) ==
sizeof(int) == 32 bits in at least one compiler.

Now we're moving into a Unicode word, text is once again not just an array
of char, so we can probably expect some interesting UTF-8 related CPU
extensions at some point.

> And the stupid business with making sure numbers were aligned on
> boundaries carried on for a loooooong time after that (horrible, nasty 68K
> processors).

The 68020 onwards don't have that restriction.



More information about the london.pm mailing list