On Wednesday 09 July 2008 12:03:19 chromatic wrote:
> On Wednesday 09 July 2008 08:27:35 Andrew Johnson wrote:
> > Confirmed, those warnings have gone. I'm still getting loads of
> > "warning: statement not reached" but I'll work out how to suppress those
> > and post a fix separately.
>
> Does your compiler have a function annotation which means "This function
> does not return"? If so, we can add a probe and a define for that.
Sun documents the use of a #pragma for that kind of annotation, which I don't
think will be compatible with Parrot's usage since I don't think you can use
C macros to declare pragmas. Here are their examples from the manual:
> extern void exit(int);
> #pragma does_not_return(exit)
>
> extern void __assert(int);
> #pragma does_not_return(__assert)
However I suspect it also understands the /[EMAIL PROTECTED]@*/ comment that
the
PARROT_DOES_NOT_RETURN macro adds since the vast majority (130) of the
warning lines I see are identical:
"lib/Parrot/Pmc2c/PCCMETHOD.pm", line 465: warning: statement not reached
That line of PCCMETHOD.pm is the one that emits "goto no_return;".
BTW, there is a fundamental problem with this very common usage of a
Parrot::Pmc2c::Emitter:
$e->emit( <<"END", __FILE__, __LINE__ + 1 );
... code ...
END
In the above usage, __LINE__ returns the line number of the line containing
the "END" marker, not that of the line containing the "emit" call, so the
#line directive it emits points to the wrong place in the source file.
The other warnings I get on Solaris are just these:
"src/stm/backend.c", line 969: warning: statement not reached
"./src/pmc/nci.pmc", line 486: warning: statement not reached
"src/encodings/utf16.c", line 326: warning: statement not reached
"src/encodings/utf16.c", line 455: warning: statement not reached
"src/encodings/utf16.c", line 479: warning: statement not reached
"src/encodings/utf16.c", line 502: warning: statement not reached
"src/encodings/utf16.c", line 525: warning: statement not reached
These all look to be valid warnings — the first is a return statement
immediately following a call to real_exception(), and that's why I think the
Sun compiler understands the /[EMAIL PROTECTED]@*/ comment, since it knew
before I
did that real_exception() doesn't return. The utf16 warnings are all from
functions that contain UNIMPL; followed by one or more UNUSED() markers.
- Andrew
--
Talk is cheap. Show me the code. -- Linus Torvalds
