[ZODB-Dev] Revisiting the ConflictError patch

[Greg Ward]

OK, this is weird.

If I check out a clean StandaloneZODB, build with Python 2.1, and run
the test script, all is well.

If I apply the first part of my patch (POSException.patch, which beefs
up ConflictError and adds ReadConflictError, but touches nothing else),
then Python dumps core while running the test script.

Wot the ... ?$$?!?!?$

Here's the output of "python test.py -vv":

  $ python test.py -vv
  BTrees.tests.testBTrees 16
  testBigInsert (BTrees.tests.testBTrees.TestIOSets) ... ok
  testClear (BTrees.tests.testBTrees.TestIOSets) ... ok
  [...eliding ~300 successful tests...]
  testValuesWorks (BTrees.tests.testBTrees.TestIIBTrees) ... ok

  ----------------------------------------------------------------------
  Ran 313 tests in 13.871s

  OK
  BTrees.tests.testConflict 16
  testFailMergeDelete (BTrees.tests.testConflict.TestIOSets) ... zsh: segmentation fault (core dumped)  python test.py -vv

The C traceback is much as I reported last time around.

Repeat the whole exercise with Python 2.1.1: same result, Python dumps
core when I run the test with my modifications to ZODB/POSException.py.

I'll keep digging...

        Greg