[Xpilot-hacks] two server crashing bugs

[William Suetholz]

Hello,
  Which distribution do you control?  There seems to be 3 distinct
distributions going on..  None of which do everything..

  I have modifications to the client code in the code from
xpilot-ng and xpilot, to allow for joystick control under Linux..
Not quite ready for prime time because of issues on lagged sites,
but I can provide patches of what I have so far, if there is any
interest, in providing at least some support for that.  I have mostly
been running the Poly client.  I have not ported my changes to
Windows or BSD.  Possibility exists, that it would be better
done using a 3rd party library.  I looked at a couple, but didn't
like the way I'd have to interface them.  One of them was a C++
library, that wouldn't fit all that well with the C coding presently
in use.  (and no, I don't want another war about the C++ version
vs the C version) :-0.

  I understand that there was some discussion going on about
the possibility of a server mod to allow sending the absolute
heading rather than a delta from current heading, which is what
I've had to deal with.  I don't know how far that has gone.
Still would have to worry about dropped packets though.
I also understand that some have felt that that kind of a patch
would allow auto pilot hacks to be easier..  I don't really follow
that, because, you can do an autopilot with the current delta
based code just as easily.


  I'd really like to see the Poly and Non-Poly servers combined,
so that the code bases would be a bit more simplified..
AFAIK, the Poly code is missing wormholes, and then it would
be able to become code base to go forward with.  I haven't
personally run across anything else it's missing.

Bill

----- Original Message ----- 
From: "Bert Gijsbers" <bert at xpilot.org>
To: <xpilot-hacks at nslug.ns.ca>
Sent: Tuesday, September 16, 2003 4:16 PM
Subject: Re: [Xpilot-hacks] two server crashing bugs


> Kristian was kind enough to email me about his bugfix patch personally.
> To me the bugs look like the kind we wouldn't want to keep in XPilot
> distributions. Therefore I propose to make a new release.  Sofar it
> includes only Kristian's bugfix patch. Are there other things which should
> go into a new release? If so please include my email in the CC list.
>
> greetings,
>
> Bert
>
> On Sat, Sep 13, 2003 at 12:41:23PM +0300, Kristian Söderblom wrote:
> >
> > On Thu, 4 Sep 2003, Kristian Söderblom wrote:
> >
> > > The other bug is in the rfrac() function. The function casts its
> > > return value to DFLOAT (float on Unix-like systems), potentially
making
> > > the value equal to 1.0 (this happens if randomMT() returns a value
greater
> > > or equal to 0xffffff80). The problem with this is that rfrac() is
> > > often used to obtain a random index for an array, gradually leading to
> > > server memory corruption, and a crash.
> >
> > Hi.
> > Jean-Yves Lefort (Decadix) informed me that the patch I've made
> > for XPilot 4.5.4 did not work as intended. The problem with it was that
> > the rfrac() function was changed to return double, but the new prototype
> > for rfrac() was not included in some c-files, which caused the compiled
> > code to malfunction.
> >
> > I've made a new patch for XPilot 4.5.4, which fixes the "get" and
> > the "rfrac" bugs and a few other minor bugs and compiler warnings.
> >
> > http://xpilot.sourceforge.net/patches/454fix.diff
> >
> > Sorry for any inconvenience,
> > Kristian Söderblom
> >
> _______________________________________________
> Xpilot-hacks mailing list
> Xpilot-hacks at nslug.ns.ca
> http://nslug.ns.ca/cgi-bin/mailman/listinfo/xpilot-hacks
>