64-bit warning cleanliness patch

[AnonymousHero]

I've attached a little patch to cmd4.c to get it to compile cleanly on 64-bit (Ubuntu). The patch is against r1867.

[konijn_]

I've attached a little patch to cmd4.c to get it to compile cleanly on 64-bit (Ubuntu). The patch is against r1867.

Thank you. It would seem that there has to be a better way than declaring those numbers , no ?

T.

[AnonymousHero]

Yeah, it seems a bit excessive, but I couldn't think of a prettier (and portable) way to do it.

[zaimoni]

Thank you. It would seem that there has to be a better way than declaring those numbers , no ?

T.

One-line fix for C99 compilers (including GCC on Ubuntu):

-       int page = (int)vpage;
+        intptr_t page = (intptr_t)vpage);

Casting void* to intptr_t is always safe in C99. MSVC doesn't have stdint.h as it only implements C90, so the above isn't usable as-is.

Note that on sufficiently strict hardware, you die on the function call to do_cmd_option_aux anyway. If the hardware both requires special pointer registers to do anything, and faults instantly on an invalid pointer entering those registers: fatality during evaluating the arguments for the function call, if not in constructing the reference data structs to begin with.

So, if MSVC is to be supported, the given patch is indeed minimal.

[Pete Mack]

A reasonable thing to do is use size_t (or long), and to typedef size_t appropriately for windows on x86 and x64.

Also, can't you use 4L etc for long integers?

[zaimoni]

A reasonable thing to do is use size_t (or long), and to typedef size_t appropriately for windows on x86 and x64.

#include <stddef.h>

(which is already happening, and will typedef size_t even in C90) Unfortunately, it's implementation-defined in C99 whether size_t is the same as uintptr_t.

Also, can't you use 4L etc for long integers?

Yes. That's probably not what's frying 64-bit *NIX, however.