Re: [AD] Current SVN OS X issues

[Evert Glebbeek <eglebbk@xxxxxxxxxx>]

On 28 Jan 2010, at 3:28 , Peter Hull wrote:
> Right, I believe that it analysed the release code, therefore those
> asserts will be NOPs.

Ah, that could explain it. I have Allegro configured for debug mode.

> So technically it's right, there is potential
> null pointer dereference but it's not very helpful to us. I don't know
> if there's a way to mark the source with a comment to tell clang "yes
> I know about this" - because maybe one of those 246 "errors" is real
> (I haven't the patience to go through them!)
> I used clang/llvm 2.6, the released version from llvm.org.


Ok, I seem to have Clang 2.4, but I think the analyser is a separate package?

> Yes, I was in the process of improving the joystick/gamepad code
> because the current one is too simplistic (and contained a bug as you
> discovered recently) but I really don't have time to sit down and do
> it properly. I forgot to clear my changes out and do an svn revert
> before I ran the analyzer, sorry.

No problem.

> I can try and document what I was
> thinking and store it as a patch on sf.net, maybe, if that would help.

If it works, sure.
To be honest though, I'm not likely to sit down, merge the patch in and test it any time soon.
The current implementation may be limited (is the worst thing about it only detecting one axis?), but at least it seems to work now. From my perspective there is no immediate rush to replace it, so we might as well leave it for when you do find some time to look at it.

Evert