[bug #27008] NSNextMapEnumeratorPair() in NSConcreteMapTable treats C-types as objects

[Richard Frith-Macdonald]

Follow-up Comment #3, bug #27008 (project gnustep):

I think you have misdiagnosed the problem ... but you can be forgiven for
doing that since the code is horribly obscure due to the way things were
hacked when Apple decided to change map and hash tables when they added new
APIs and classes while retaining the old C APIs but suing the same names for
both old and new.

I've added some comments to the source code about this ...
If the 'map' field of the enumerator is nil, the enumerator is not a
GSIMapEnumerator and contains objects in the locations normally using for the
node and bucket.

The thing is, this should really only be the case if you have subclassed the
new NSMapTable class and are using an enumerator on that subclass.  You may
well be doing that, but it would be unusual, so I suspect you aren't.
That would lead me to guess that your NSMapEnumerator is corrupt ... eg an
uninitialised variable, but that's pretty much a stab in the dark.

Perhaps you could supply a small test program demonstrating the problem ...
then I could have try at running it to reproduce the issue, and examine things
under gdb to see what's going wrong.


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?27008>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/