Follow-up Comment #3, bug #23781 (project freeciv):
There are some other a bit similar problems, and when thinking these as a
whole, the most sensible semantics for this would probably be to consider "No
One" as one we have "No Contact". This does not remove the fact that the
low-level requirement code can't know if NULL means "don't know" or "no one",
so fake player to whom nobody would have a contact could be one solution, but
rather ugly one (There would probably be hundreds of places where the code
breaks when the seemingly valid player pointer actually isn't - or
alternatively specific out-of-bounds player pointer should be accepted as "No
One"
> I'm not familiar enough with the users of RPT_POSSIBLE to tell
> without doing a code review.
If that would cause problems to some caller, it's broken already. If you don't
ask semantics of CERTAIN, you should not expect semantics of CERTAIN, i.e.
when ever you ask POSSIBLE you may get a false positive (with CERTAIN you may
get false negative)
_______________________________________________________
Reply to this item at:
<http://gna.org/bugs/?23781>
_______________________________________________
Message sent via/by Gna!
http://gna.org/
_______________________________________________
Freeciv-dev mailing list
[email protected]
https://mail.gna.org/listinfo/freeciv-dev
