Messages in this thread |  | | From | Peter Samuelson <> | Date | Tue, 17 Oct 2000 06:22:14 -0500 (CDT) | Subject | Re: [Criticism] On the discussion about C++ modules |
| |
[Peter Samuelson] > > Is this similar to the gcc 'const' attribute? > > > > int foo (int, char *) __attribute__((__const__)); > > > > This is valid in GNU C (not just C++). Read the info page for > > details.
[Eray Ozkural] > Probably, I haven't used it in my C code though. I've found an > article explaining the C9x restrict keyword at URL > > http://www.cuj.com/archive/1707/feature.html
OK, it appears your 'restrict' keyword has *nothing* to do with "constant functions" as gcc implements them. 'restrict' is all about pointers not being aliased. A gcc constant function, OTOH, is one guaranteed not to have any side effects, so its return value can be cached and reused by the optimizer. Two completely different things.
And I have no idea if gcc implements 'restrict' or not. As of 2.95 it does do type-based alias optimizing, so I imagine at least a lot of the infrastructure is there.
Peter - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/
|  |