On Sun, Jan 17, 2021 at 10:36 AM Greg Ewing <greg.ew...@canterbury.ac.nz> wrote: > > On 16/01/21 4:17 pm, Chris Angelico wrote: > > But somewhere along the way, you're finding that there's a problem, > > which implies that SOMETHING is calling on C stdio. That thing, > > surely, should be the thing responsible for flushing? > > The C library using stdio has no way of knowing that something > else (e.g. Python) is bypassing stdio. That's an unusual thing > to do in the C world, so you can't really blame the authors > of the library for failing to anticipate it. > > It's also unreasonable to expect a C library to flush after > every write, as that would defeat the purpose of stdio buffering. >
This is true. However, at some point, the boundary is crossed from Python into the C library. Something, at that point, knows. It's very common to have a flush option available, so it should be used. It's extremely uncommon for Python to call into C code that (a) uses C stdio, (b) shares the file handle with something else (eg stdout/stderr), and (c) doesn't have an option to flush its output. And when that happens, I would consider it a bug, not just because of the shared handle - if you're piping the output into something else, you might need to force it to flush its buffer. ChrisA -- https://mail.python.org/mailman/listinfo/python-list
