[ODE] Warnings and Errors ....

[Mark Riedesel]

On Mon, 27 Sep 2004 11:53:27 -0700, Jon Watte <hplus-ode at mindcontrol.org> wrote:
> 
> > > > the ones who choose to ignore warnings
> > > > can continue to do so and those who don't can start to fix them
> 
> > > No, because those who ignore warnings have commit privileges,
> > > and those who care about them, don't.
> 
> > 2.  You don't run this one.
> 
> That's exactly what I said.
> 
> > 3.  Your (?) patches to fix the WARNINGS WHICH ARE NOT ERRORS that you've
> > found are going to be committed, if they haven't been already.
> 
> That hasn't been my experience. In fact, in standard ODE, you
> can still push ccylinders through boxes because of divisions by
> denormals causing infinities -- my patch was rejected on stylistic
> reasons alone, yet you imply that things that BREAK THE COMPILE for
> a number of users are OK, for stylistic reasons.
> 
> In fact, one of the problems I pointed out would break the build
> for ANYONE using dReal==float.
> 
> My post was an attempt to make the maintainers realize that
> cleanliness of compile is really important to some members. It
> appears they don't care, and that's the end of the discussion.
> As I said last time: I said my bit, I'm not going to change
> anyone's religion, so I'd appreciate it if you toned down the
> personal attacks.

Not meaning to be an antagonist or anything (although I probably am),
but I've always understood that using the CVS version of anything
involves a risk of broken compiles and/or really freaky bugs, which
makes sense why most projects have a "latest stable release", which in
the case of ODE is 0.5

Although I do agree this whole debate needs to end, I just don't like
seeing people getting upset when code (which is in testing, and heavy
development) suddenly breaks, I do it to my own projects all the time.