[Python-Dev] Preventing 1.5 extensions crashing under 1.6/2.0 Python

[Mark Hammond]

> Wasn't there a problem with this, because the 'Py_FatalError()'
> would be the
> one in the uninitialized library and thus result in the same
> tstate error ?
> Perhaps it needs a separate error message, that avoids the usual Python
> cleanup and trickery and just prints the error message and exits ?

I would obviously need to test this, but a cursory look at Py_FatalError()
implies it does not touch the thread lock - simply an fprintf, and an
abort() (and for debug builds on Windows, an offer to break into the
debugger)

Regardless, I'm looking for a comment on the concept, and I will make sure
that whatever I do actually works ;-)

Mark.