[hawkinsw]

I've written lots of servers in C and found this article quite interesting.

The techniques that the author presents are new to me and I know that I will take them into consideration when I write a server in C the next time.

More importantly, they are techniques that I think library writers should use. My uneducated guess is that the underlying implementations of servers in Node.js and Iron, etc, could be improved with these ideas, even though they might not be directly applicable since those servers are written in different languages.

I can't vouch for the results, but the author's experiments seem conclusive.

[DSMan195276]

His allocating an fd_set to an arbitrary size is somewhat iffy standards wise, but if it works it works - especially for a solution for specific OSs like Linux, where this won't be broken in the future. On that note, it is worth wondering how expensive the copying is, but I would wager he is right in that the copying isn't actually that expensive - it's only an 8K buffer.

Also, when creating an array that is indexed by fd, if you want to only use the memory you need instead of creating an extremely large array for the max fd of the system you can just allocate the memory using `mmap`. Systems like Linux won't actually allocate all the memory right when you do the `mmap`, instead it will map in a zero-page in each spot. When you write to it, the kernel will automatically allocate a page to back that memory. When you combine that with the kernel always selecting the lowest fd, it means that the memory backing your array will only grow to be as large as needed for the maximum number of fd's you use. (Note: It's likely that such a large array would be allocated via `mmap` in it's own spot by malloc/calloc anyway, but using `mmap` ensures you get the functionality you intended.)