[SDL] SDL_Net Server Stuff
Steve Salkin
steve at salkin.org
Sun Feb 17 21:59:01 PST 2002
Nicolai Haehnle <prefect_ at gmx.net> wrote:
>> Big guru R.W.Stevens says that connected UDP sockets have
>> significantly less overhead (the kernel seems to connect implicitly
>> for each sendto which can be avoided by this). (But maybe that's not
>> a problem with winsock and maybe current Linux has solved this, after
>> all, the books have been written years ago? Someone knows more?)
>>
>> He recommends having one socket per connection. So I would think:
[snip]
> Ugh. I don't know who R.W.Stevens is, but logic tells me that he's
> just plain wrong.
> First of all, if there is any overhead with unconnected UDP sockets,
> it seems to be a problem in one particular kernel, and as you say
> yourself should be eliminated by now. This problem may or may not
> exist (it's likely to not exist I should think) in other kernels.
> Since SDL is all about writing portable code, you shouldn't optimize
> for a single crappy system, especially when this "optimization"
> results in a general decrease of efficiency.
He's talking about W. Richard Stevens, the late author of numerous books
on network programming and advanced programming under Unix and winner of
the USENIX Lifetime Acheivement Award. In his lifetime he probably
forgot more about programming in general, and network programming
specifically, than you are ever going to know.
WRS Home Page: http://www.kohala.com/start/
WRS Obituary:
http://dan.yosemite.ca.us/6bone/w.richard.stevens.obituary.html
That said, I don't think that Mr. Burstedde, the original poster,
accurately represents Mr. Steven's position, which was hardly so
unequivocal.
S-
More information about the SDL
mailing list