[Freeciv-Dev] Re: Crash in 7-27-2001 CVS GTK+ client (PR#869)

[Kevin Brown <kevin@xxxxxxxxxxxxxx>]

Gaute B Strokkenes <gs234@xxxxxxxxx> wrote:
> On Sun, 29 Jul 2001, kevin@xxxxxxxxxxxxxx wrote:
> > 
> > NDEBUG is set by the configure script unless --enable-debug is
> > passed to it.
> 
> This is wrong; NDEBUG is set iff you explicitly turn it off.

Ah, okay.  You're right.  The RPM spec file uses --enable-debug=no.
Didn't notice that before.  It was late last night (if you couldn't
tell) that I looked at it.  Sorry.

Even so, one should NEVER put code that has side effects in an
assert()!!  So I think my original stance remains: someone needs to go
through the code and fix all the assert()s so that they don't call any
code that has side effects.

I guess I could take on that job if nobody else wants to, though I
don't have CVS commit access and I didn't see any way of getting a
jitterbug account (perhaps they're related?).

> > So on any non-debugging build (like the rpm install I do), assert()s
> > will translate to no-ops and, in this specific case,
> > normalize_map_pos() will never be called.
> 
> Did you do a RPM install of a CVS version?  How?

Took a spec file that worked for 1.11.4 and modified it appropriately.
I'll send you the spec file if you're interested, or put it on the
list if others want me to.



-- 
Kevin Brown                                           kevin@xxxxxxxxxxxxxx

    It's really hard to define what "unexpected behavior" means when you're
                       talking about Windows.