Marc Culler

I think I found the explanation for this problem in the linux termios man page where it explains the TCSAFLUSH flag to tcsetattr: > the change occurs after all output...

This is almost surely a cache problem, which is very common with the jupyter server. (See [this SO page](https://stackoverflow.com/questions/55152948/juypter-notebook-shows-blank-page) for example.) Changing the host name from localhost to 127.0.0.1, or...

I don't think this is a bug. It is not exactly a feature, but it is a consequence of something which is regarded as a feature. It is also not...

I retract my statement that this is not a bug. By adding some print statements in the Gen `__dealloc__` method I was able to demonstrate that there are lots of...

An addendum. While the refcount behavior above looks wrong for the pari int Gens, it does not cause the same kind of memory issues to replace `M = pari.matrix(10, 10,...

If you think through what happens when you repeatedly call `M = pari.matrix(10, 10, range(100))` the behavior we are seeing is exactly what you would expect. At the beginning of...

If what I said above is correct, it is definitely not the whole story since it does not account for the fact that many Gens are moved to the heap...

I can now finish the story. The discussion above explains why the loop `>>> while True:` `... M = pari.matrix(10, 10, range(100))` generates many matrices on the pari stack. The...

Oh, and the fix is to call A.fixGen() as soon as a matrix A is created.

@videlec @NathanDunfield A simple implementation of this idea might be to just have a setup.py option which specifies whether to use system libraries or to build the libraries from scratch....