Tian Gao

The current basic test fails on Windows and I'd love to leave this bug to you guys as a first issue to work on!

> The document doesn't seems to illustrate the save format of other event, e.g. instant event. Is this determined by me? I would suggest to make a PR the the...

> > For instant event, I consider using the following format: header(header) - pid(pid) - tid(tid) - name(str) - count(uint64) - [start(ts) - scopes(str)]* > > header(header) - pid(pid) -...

> > Should probably get a better idea of instant events. For example, `args` is a critical part of an instant event which contains the data that user logs. >...

I believe this is a ``numpy`` specific issue. I did not dig too much into it but I believe ``numpy`` has some optimization there to prevent the python profile hook...

I agree that this is a very important usage. ``numpy`` probably has some optimization that made the ``add`` function a pure C function, instead of a PyCFunction. I'm not an...

@ogrisel please notice that py-spy is a statistical profiler so it's impossible by nature to have a "synchronized timeline" between threads. Many people are kind of confused(and it's completely normal)...

@AndreKR Every advantage you said about sampling profiler is true. And yes, you can get the full stack of every thread at the same time, you can even put them...

Yes. Sampling profiler can provide timing information about the function that's significantly larger than sampling period. In your example, if you can "cache" the template, then sampling profiler is definitely...

That's right. Let's steer back to the topic(of which other people might have better ideas).