Re: Heap usage and variable argument C functions

[David Given <dg@...>]

On Tuesday 21 December 2004 12:25, Brett Bibby wrote:
[...]
> I didn't say it was illegal, that's why it's a warning and not an error.

I'm sorry, I misspoke: I didn't mean legal, I meant *expected*. It would not 
occur to me that putting a long into an int would produce an warning because 
that would not strike me as being worthy of a warning except in special 
circumstances (such as the compilers people are mentioning).

*shrug* I'm rather coddled; most of the platforms I've been dealing with are 
gcc-based, which means I get very used to a certain type of behaviour. (I 
remember once trying to compile a port of Nucleus with gcc instead of the 
proprietry compiler that it had been written with, and discovering that none 
of the device drivers worked: the proprietry compiler treated all variables 
as volatile, which gcc didn't, and the original author hadn't bothered to 
mark any of the hardware registers as volatile, so of course gcc was 
optimising out all the delay loops.)

But it does seem to me that you have to write all code with a set of 
assumptions in mind, and Lua's assumptions are that it's going to run on a 
standard C system. If your system has different assumptions or requirements, 
then I think it's reasonable that you should be expected to make 
modifications. So I don't think it's fair to raise that issue as an actual 
Lua *bug*.

That said, if you were to submit a patch, I'm sure the maintainers would be 
interested in considering it as a useful new *feature* (i.e. compiles cleanly 
on systems with non-standard behaviours X, Y and Z).

-- 
+- David Given --McQ-+ "A psychic? That sounds like science fiction."
|  dg@cowlark.com    | "Dear, you live on a spaceship."
| (dg@tao-group.com) | "So?"
+- www.cowlark.com --+ --- Firefly, _Objects in Space_