Dima Pasechnik

> For me, the symptom is sporadic. If the viewer opens up the file quickly enough, no problem. Sometimes not. Maybe dependent on the pdf file size... Do you by...

> No. > > I explained how it happens above: [#38339 (comment)](https://github.com/sagemath/sage/pull/38339#issuecomment-2267958420) > > There is nothing strange why it happens. Do you mean it's a bug in CPython on...

reading `https://docs.python.org/3/library/threading.html` does not convince me about the validity of the syncronisation via `t.daemon` used in Sage. It seems to be for something else.

it seems to be the default behaviour of `open` on macOS is not to wait until the process ends. However, one can do `open -W` to let it wait until...

the proper way to have temporary files cleaned up is to get them registered with sage-cleaner using `atexit`, see `sage.cpython.atexit`. @orlitzky knows all about it :-)

Do all these viewers really load the whole pdf file into RAM - or we are just lucky to have examples fit into an i/o buffer? The only safe solution...

> Anything else? > > The problem of not cleaning up the temporary file if sage session ends while the viewer is open persists, before and after the PR. This...

I checked that this builds if merged over the latest beta. Could you fix "This branch is out-of-date with the base branch" ? The interactive rebase feature as provided by...

These rebases unbreak the CI - so that's the main reason to do it.

Mathematically it's not echelon form, and not even Gram-Schmid, as you have a twist by the Galois automorphism in the factorisation.