Re: C helloworld from inside Ravi

[Dibyendu Majumdar <mobile@...>]

Hi Sean,

On 20 June 2017 at 01:13, Sean Conner <sean@conman.org> wrote:
>   That's something my own C compiler module [1] does:
>
>         local cc = require "org.conman.cc"
>
>         local CODE = [[
>         #include <stdio.h>
>         #include <lua.h>
>
>         int TestFunc(lua_State *L)
>         {
>           puts("hello world!\n");
>           return 0;
>         }
>         ]]
>
>         local f = cc.compile("TestFunc",CODE)
>         f()
>
>   -spc
>

How do you handle following code:

local testfunc = [[
struct lua_State;
extern int puts(const char *);
extern int TestFunc(void);

int TestFunc(void)
{
puts("hello world!\n");
return 0;
}
]]

I try to detect that the function has the expected prototype so that
above is rejected as below.

..\llvm64d\Debug\ravi: dmrc_test.lua:17: bad argument #1 to
'getfunction' (Invalid function: canot accept more than one argument
)
stack traceback:
        [C]: in method 'getfunction'
        dmrc_test.lua:17: in main chunk
        [C]: in ?

Error message can be improved of course !

Anyway the issue with allowing users to write C code is basically that
it becomes very easy to crash the system. This is why I have not been
keen on providing automated FFI support in Ravi.

Regards
Dibyendu