Cody Piersall

There is a semantic difference: now the callback is called *with* the `s_pipe_cbs_mtx` held. I suppose this means that users cannot change callbacks within a callback on pain of a...

> Is a mutex strictly necessary here? I see your point, @bb010g. This patch is using the mutex for more than what mutexes are traditionally used for--protecting data structures from...

I may have been wrong about this being an issue in nng. The following C code I expected to reproduce this without using the Python bindings, but I couldn't reproduce...

Ugh, I realized that in my "reproducer" above, I was potentially watching pipe events from the wrong socket, if the short-lived sockets cause a race in the longer lived ones....

Okay, it took forever for me to get back to this, but **I have finally made a plain C reproducer** that demonstrates the bad pipe event ordering, and it is...

@benfred What do you think about changing the default behavior to be as if `--threads`` were given? The biggest drawback I can think of is that people may have come...

If py-spy is using the Python ABI (instead of API), then for Python >= 3.8, this won't be a problem, because the ABI between release and debug is the same....

Whoa, Python back to v2.3! I _did not_ realize that. Cool!

It's the cumulative number of deaths in the county as of the date in the same row. So each county will have multiple entries, one per day.

This is an old issue, but when I had this problem it was because I didn't have nanomsg header files on my computer. I just had to install them. If...