On Fri, 2007-04-13 at 15:20 +0100, Rhythmic Fistman wrote:
> On 4/13/07, skaller <[EMAIL PROTECTED]> wrote:
>
> > Signed types don't support bitwise operations
>
> Why not?
The semantics are platform dependent in C: in C89 the
semantics are indeterminate.
The C90 standard has words that make it appear otherwise
but the words weren't written by people that knew any
maths .. they're total gibberish.
In C99 its better, but representations are mow overconstrained,
due to an idiot who didn't know any maths (or much about C) ..
but still not enough for bitwise operations to be
deterministic on signed values.
Felix mandates (somewhere .. ) two's complement,
so the operations would be deterministic, nevertheless
I think it is very bad practice so it isn't supported
in the library, deliberately. You can of course
roll your own or argue to change the library:
my choice is initial not terminal :)
Roughly speaking signed integers are intended as
local (close to 0) approximations to integers,
and any other use is incorrect.
Note the shift operators aren't bitwise operators,
they're binary exponentials (this is true in C99 too).
If you want to twiddle bits, unsigned values are
both bit strings and also modular values: all the
operations are total and platform independent
(modulo the number of bits).
> Or why does int -> uint require an explicit cast?
Because ALL conversions in Felix require an explicit cast.
Well actually this isn't quite true.
1. Lvalue -> rvalue decay is implicit
2. Function class -> closure is implicit
So if you see:
1 + 2u
working .. it isn't because of an implicit conversion,
it's because there is an overload on + to support
all combinations of integral addition.
I actually don't like that, but it is quite necessary
for binding libraries such as OpenGl which use crud like
GLint
which is an alias to some integer, but you don't know
which one.
if you want to set flags in signed values you can do:
int$ (uint x \| 0xF) \& 0xF0
i.e. cast to unsigned, mess about, and cast back.
If this is really too messy it can be fixed .. the closer
we get to the mess C made, the less safety advantages
Felix provides .. but compromises are necessary: binding
C library code in a fairly convenient manner is essential.
But there's usually more than one way to skin a cat ..
--
John Skaller <skaller at users dot sf dot net>
Felix, successor to C++: http://felix.sf.net
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Felix-language mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/felix-language
