[Freeciv-Dev] Re: (PR#8547) GCC poisoning v2

["Per I. Mathisen" <per@xxxxxxxxxxx>]

<URL: http://rt.freeciv.org/Ticket/Display.html?id=8547 >

On Tue, 20 Apr 2004, Raimar Falke wrote:
> Other question (which I also asked the last time): what are the
> results? Do we have such bad function usage? Why do we have to add the
> test framework into the normal code? Just add the results is fine.

I found about 4-7 erronous cases each time, I think. So it is not very
common that these mistakes are made, but when they are, this is just about
the only way that they are found ;)

Yes, we don't really have to have this activated all the time. If we just
put in all the necessary changes (moving stuff around, headers), then the
gcc poisoning _can_ be there but not used until we want to. However, if
there are no good reasons against, I think adding gcc poisoning all the
time might be just as good - it makes patch authors learn. Maybe commit
with poisoning default on, then turn it off if people get problems?

  - Per