10 Aug, 2010, Barm wrote in the 1st comment:
Votes: 0
At some point… I intended to port my telnet lib from poll to epoll and reap massive increases in speed and client juggling (on Linux) but reading this blog post makes me sad:

http://sheddingbikes.com/posts/128082938...

That's Zed Shaw, who famously left the Ruby community for Python, and the Pythonista's still aren't sure if that's a win or not. It is nice to see someone take a scientific wack at the issue and in one spot he says:

Quote
Another way to put this is if your server or protocol is the type that transmits a ton of data and can't tolerate idle resources, then poll will win. This would be the case for data servers, media servers, streaming data, image caches, anything where your goal is to pump data, lots of it, and get it out fast. In this case that's because your active/total ratio is greater than 0.6 most of the time.

But, if your server is the kind with lots of idle connections just hanging out, then epoll wins not poll. That's because your active/total ratio is < 0.6 most of the time.


Where MUDs would clearly fall into the second category. Still, I've been looking at epoll as an untapped panacea – which apparently, it aint.
28 Aug, 2010, Barm wrote in the 2nd comment:
Votes: 0
A followup:

As I mentioned in another thread, I've been re-writing my single threaded, asynchronous library. I have separate Epoll() and Select() servers depending on the user's OS. I've also written a small HTTP service that runs off this lib. Currently, it only handles GET requests.

To test performance, I'm using a HTTP benchmarking utility called siege.

Here's my test command:
$ siege -b -c 1 -r 2000 http://127.0.0.1:7777


Here's the results when using Select():
Transactions:		        2000 hits
Availability: 100.00 %
Elapsed time: 1.40 secs
Data transferred: 0.39 MB
Response time: 0.00 secs
Transaction rate: 1428.57 trans/sec
Throughput: 0.28 MB/sec
Concurrency: 0.95
Successful transactions: 2000
Failed transactions: 0
Longest transaction: 0.01
Shortest transaction: 0.00


And here's the results when using Epoll():
Transactions:		        2000 hits
Availability: 100.00 %
Elapsed time: 1.31 secs
Data transferred: 0.39 MB
Response time: 0.00 secs
Transaction rate: 1526.72 trans/sec
Throughput: 0.29 MB/sec
Concurrency: 0.94
Successful transactions: 2000
Failed transactions: 0
Longest transaction: 0.01
Shortest transaction: 0.00


The benchmark was run on an older, single core, 2.2ghz Athlon64. The results show both methods really close with a tiny advantage going to Epoll(). This test is rather specific because we're accepting a connection, stuffing the requested file (index.html) down the socket, and dropping the connection once it's cleared. A test of persistent connections might be more revealing.
29 Aug, 2010, Barm wrote in the 3rd comment:
Votes: 0
For the record, the title of this thread is "Damn you, Zed Shaw". I noticed that if you're not logged on it's reads **** You, Zed Shaw which of course looks like F*** You. Apologies to Mr. Shaw. I wish I had used 'Darn' now – can a mod change it?
29 Aug, 2010, tphegley wrote in the 4th comment:
Votes: 0
Barm said:
For the record, the title of this thread is "Damn you, Zed Shaw". I noticed that if you're not logged on it's reads **** You, Zed Shaw which of course looks like F*** You. Apologies to Mr. Shaw. I wish I had used 'Darn' now – can a mod change it?

Done
29 Aug, 2010, Tyche wrote in the 5th comment:
Votes: 0
Barm said:
For the record, the title of this thread is "Damn you, Zed Shaw". I noticed that if you're not logged on it's reads **** You, Zed Shaw which of course looks like F*** You. Apologies to Mr. Shaw. I wish I had used 'Darn' now – can a mod change it?


Why would anyone apologize to him? Let's just expose this wannabe for what he is. He's so completely and totally useless, stupid, and arrogant, one wonders why anyone would bother linking to anything he wrote.

Post edited by Dean.
29 Aug, 2010, Barm wrote in the 6th comment:
Votes: 0
I must be missing some context here, Tyche. I only know him from the "Rails is a Ghetto" rant and his slagging of Mark Pilgrim's Dive into Python (which was a bit tacky) but his test of Epoll vs. Select seemed reasonable. Why the hate? Not defending him – I half expect to hear "Python is Ghetto" as he moves on to Lisp.
29 Aug, 2010, Rudha wrote in the 7th comment:
Votes: 0
Ahhh … LISP. The one language I actually enjoyed learning in CS classes, and consequentally doomed to never be used in the Real World TM As long as there have been computers, there have been holywars over certain tools being better than others. I still to this day hear arguments of EMACS versus VIM.

Anyways, Im not sure what we're trying to prove here. That someone says something is better? That's objectively testable and I believe that was done above. So … where are we going here? I'm confused.

Maya/Rudha
30 Aug, 2010, David Haley wrote in the 8th comment:
Votes: 0
What we're seeing in any case is that a post like that is apparently acceptable to MB mods…
30 Aug, 2010, Ssolvarain wrote in the 9th comment:
Votes: 0
Tyche said:
Barm said:
For the record, the title of this thread is "Damn you, Paris Hilton". I noticed that if you're not logged on it's reads **** You, Lindasy Lohan which of course looks like F*** You. Apologies to Miley Cyrus. I wish I had used 'Darn' now – can a mod change it?


Why would anyone apologize to him? Let's just expose this wannabe for what he is. He's so completely and totally useless, stupid, and arrogant, one wonders why anyone would bother linking to anything he wrote.


See, it's a matter of context. If I replaced the names with "David Haley" it would be different.
30 Aug, 2010, David Haley wrote in the 10th comment:
Votes: 0
It has been fairly well accepted that blatant swearing is not ok, and the rules expressly forbid bigotry, which the homosexual slurs would fall under.
30 Aug, 2010, Tyche wrote in the 11th comment:
Votes: 0
Barm said:
Tyche said:
Why would anyone apologize to him? Let's just expose this wannabe for what he is. He's so completely and totally useless, stupid, and arrogant, one wonders why anyone would bother linking to anything he wrote.

I must be missing some context here, Tyche. I only know him from the "Rails is a Ghetto" rant…


The context you seem to be missing is Zed Shaw's "Rails is a Ghetto" rant, which I borrow liberally from above. It's not even about Rails (or Ghettos).

Now there have already been a half dozen papers observing this exact behavior of epoll and select. For starters see, "Comparing and Evaluating epoll, select, and poll Event Mechanisms" by Gammo, Brecht, Shukla, Pariag (2004). His article illustrates the same tendency to build straw men up and knock them down.
30 Aug, 2010, Cratylus wrote in the 12th comment:
Votes: 0
Tyche said:
Post edited by Dean.


Aren't user-configurable profanity filters in place to avoid the need for this mod action?

-Crat
http://lpmuds.net
30 Aug, 2010, David Haley wrote in the 13th comment:
Votes: 0
It's not so much the profanity as the bigotry, although the profanity was also rather gratuitous and more appropriate for the controversial section.
30 Aug, 2010, Tyche wrote in the 14th comment:
Votes: 0
Cratylus said:
Tyche said:
Why would anyone apologize to that scrawny cock sucker? Let's just expose this wannabe fuck for what he is. Zed Shaw is a pathetic fucking tool bag piece of shit, a little ghetto fabulous princess arrogant fuck fat fucking face wannabe fucktard loser. He's so completely and totally useless, stupid, and arrogant, one wonders why anyone would bother linking to anything he wrote.


Aren't user-configurable profanity filters in place to avoid the need for this mod action?

-Crat
http://lpmuds.net


I'm not sure what the problem with the post is, nor how it violates any of the site's rules.
30 Aug, 2010, Barm wrote in the 15th comment:
Votes: 0
Ladies, can we talk programming for a change?
30 Aug, 2010, David Haley wrote in the 16th comment:
Votes: 0
Well, Barm, you asked the question that prompted this. :shrug:
30 Aug, 2010, Rudha wrote in the 17th comment:
Votes: 0
I don't know what was the expected response, with that kind of topic title, to be honest.

Maya/Rudha
0.0/17