[Python-bugs-list] Fatal Python error: PyThreadState_Delete: invalid tstate (PR#299)

[Tim Peters]

[greg@itasoftware.com]
> I was getting this error:
>
>    Fatal Python error: PyThreadState_Delete: invalid tstate
>
> in a threaded program.
>
> Eventually I looked on your bug list pages and found the fix,
>
> ...
>
> My question is, what is some normal user supposed to do?

Well, (a) a "normal user" usually doesn't use threads at all, and (b) this
bug was exceedingly rare even among abnormal <wink> users; e.g., it took me
extreme efforts to provoke the bug at all on a uniprocessor machine, and I
had never bumped into it before making that effort.  I filed the bug report
in the bug database after it was fixed, so that people could find it without
stumbling into the Thread-SIG archives (BTW, if you're one of Python's
serious thread users, you should sign up for that list!).

Right after the fix, Guido posted a call for volunteers to put out a
critical bugfix release, but nobody responded.  A release (even a bugfix
release) is a major effort, and nobody (including me) stepped up to the
plate.  So, you see, it's entirely your fault <wink>.

> Given that the bug was fixed in June of 1999 and you didn't have a
> next release planned for at least a year,

What makes you think that the next release wasn't planned for at least a
year?  That's not my recollection.  It will have *taken* a year by the time
1.6 gets out, but that's hindsight.

> wasn't a 1.5.3 warranted for a "CRITICAL PATCH"?

Yes, I think so, and Guido did too.  Unfortunately, nobody made time to do
the work.  That's all there is to it.

Sorry for the hassle!  Alas, until someone gets paid to do Python work, you
get what volunteers are able to do in their spare time.  The good news is
that the tstate bug is the most serious lapse I can recall in nearly a
decade.