🚀 go-pugleaf

In article <4a73am$jfu@newshost.lanl.gov>
Tanmoy Bhattacharya <tanmoy@qcd.lanl.gov> writes:
>`long *i;'
>(long*)(int*)(char*) i == i

No, it may not work.  What's the problem with that?

We've seen two examples of implementations where this won't work, which are
plausible if a bit unusual.  I'm sure that i could think of more examples.

The standard says that you can convert any object pointer to "void *" and back
and get the same pointer, and "char *" is grandfathered.  I don't see how this
implies that some round-about sequence of conversions is guaranteed to work.

In particular, i don't see why a "long *" is guaranteed to be a valid "int *".
It's often true, of course, but is there anything in the standard saying this?

--
Joe Keane, professional programmer