OT: CS education gone wrong (Was: Re: TDD is BS?)

Adam D. Ruppe destructionator at gmail.com
Mon Jun 24 09:37:53 PDT 2013


On Monday, 24 June 2013 at 02:39:30 UTC, Andrei Alexandrescu 
wrote:
> so what's the way to go now? One process per request? Seems 
> heavy to me seeing as most requests last very little.

Performance wise, I've found heard very good things about async 
i/o (like vibe.d) and had very good experience with a process 
pool (on linux anyway, cgi.d's fastcgi and new 
embedded_httpd_processes do this, the latter hitting 6,000 
requests per second for hello world on my machine!). The process 
pool forks ahead of time, then they take connections as they 
become available. Basically it puts the concurrency in the 
kernel's hands, which does a good job.

The zip I linked to yesterday does a simple one thread per 
connection, multiple requests per connection model. Doesn't scale 
very well, but works well and with simplicity on light loads.



More information about the Digitalmars-d mailing list