Re: Python recompile

On 3/13/2025 4:49 PM, Chris M. Thomasson wrote:

On 3/13/2025 4:45 PM, Chris M. Thomasson wrote:

On 3/13/2025 1:36 PM, Lawrence D'Oliveiro wrote:

On Wed, 12 Mar 2025 22:25:49 -0700, Chris M. Thomasson wrote:
>

On 3/11/2025 11:50 PM, Lawrence D'Oliveiro wrote:

>
On Tue, 11 Mar 2025 21:22:22 -0700, Chris M. Thomasson wrote:
>

On 3/11/2025 8:12 PM, Lawrence D'Oliveiro wrote:
>

On Tue, 11 Mar 2025 20:04:30 -0700, Chris M. Thomasson wrote:
>

IOCP works fine ...

>
But not with pipes.

>
Never tried them with pipes, only sockets.

>
The Python developers have tried. They ended up creating two event loop
implementations on Windows, neither of which quite covers all the bases
<https://docs.python.org/3/library/asyncio-eventloop.html#event- loop-implementations>.

>
For some damn reason, I don't think you have tried IOCP with sockets,
pipes, ect, even Linux AIO?

>
If you think you know something the Python developers overlooked in
their Windows implementation, feel free to tell us about it.

>
Are you part of their development team? I have no idea how the Python dev's use IOCP with sockets. I know how I used to use it. I also don't know how many of them read that paper on the 50,000 concurrent connection contest wrt events vs IOCP. It's been many years since I have read it. There were many interesting tips and tricks... Also, there was this other paper I read on a so-called staged server design.
>
https://www.cs.cmu.edu/~StagedDB/papers/CMU-CS-02-113.pdf
>
I am having a little trouble finding the other paper...
>
Damn!

>
Actually, I am having some more thoughts it might of been in a damn book! God it been years. Back in early 2000's.

 Close but no cigar!
 https://web.archive.org/web/20060214135421/http://msdn.microsoft.com/ msdnmag/issues/1000/winsock/default.aspx
 Damn!