This is the mail archive of the
gcc-help@gcc.gnu.org
mailing list for the GCC project.
Re: Incorrect dereference in argument passage
On Wed, Sep 5, 2012 at 4:22 PM, Arthur Schwarz <aschwarz1309@att.net> wrote:
>
> I included the disassembled code in the original problem description. The entire
> coding ensemble is some 10,000SLOC. I don't know what effort or time it would be
> to construct test case to illustrate this one point. I suspect that it may take
> several days, or then again, maybe several hours. Isn't the generated code
> sufficient?
To be honest, the generated code alone tells me nothing. You are
basically asking whether the generated code is a correct compilation
of the source code. It's impossible to answer that question without
seeing the source code. You provided a verbal description of the
source code, but my sad experience is that 99% of the time somebody
believes that code has been miscompiled, it is because they have
misunderstood their source code or the language. So while you could
certainly be in the 1% with a real bug, a verbal description is
insufficient to determine that.
> I will try to generate a specific test case, but it will probably take several
> days to get "around to it". is this OK?
It doesn't matter to me one way or the other.
You did ask some specific questions in the original message:
> It looks as if in some cases using an implicit inline statement (a body and
> method definition in the header file) breaks the compiler.
This is not a known bug. If it were known, it would be extremely
serious, and it would be fixed.
> if this is accurate
> are there any guidelines?
There are no guidelines because the bug is not known.
> Will explicit inlines work correctly (inline <type>
> method();) or can inlines not be used?
Both implicit and explicit inlines can be used and people use them all the time.
Ian