| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
…e-smaller-socket-buffer
Feature smaller socket buffer
| is_pipe = stat.S_ISFIFO(mode) | ||
| limit = self._default_pipe_size if is_pipe else remaining | ||
| is_socket = stat.S_ISSOCK(mode) | ||
| limit = self._default_pipe_size if is_pipe or is_socket else remaining |
There was a problem hiding this comment.
How about use the limit for all connection, instead of socket and pipe?
Sorry, something went wrong.
There was a problem hiding this comment.
If there are no performance drawbacks for other types of connections, then it's a solid solution. I've only analyzed pipes and sockets.
Sorry, something went wrong.
There was a problem hiding this comment.
I don't know connection other than pipe and socket.
But mmap+mremap+munmap overhead is very common issue for all streams.
So using chunked read by default and override it in special connection class seems better solution.
Sorry, something went wrong.
There was a problem hiding this comment.
How about updating Misc/NEWS.d/next/C API/2024-07-03-10-11-53.gh-issue-121313.D7gARW.rst instead of adding new changelog?
Changelog is part of Python documentation. It should be readable than commitlog for Python users.
Sorry, something went wrong.
There was a problem hiding this comment.
I was prompted by the blurp it bot to create a summary of the changes via the separate website.
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
Issue: #123557