I am doing a presentation of tux, for a linux network course I take. Here are some slides I prepared if someone is interested: http://www.uni-karlsruhe.de/~uwid/tux/ they are in in Open Office format, but I exported them as web and power point as well. I have a couple of questions, mostly to Ingo and those who understand the internals, regarding Tux. I have looked at the source, but I don't understand very much. I've read Michael Tiemann's good article on dell ( http://www.dell.com/us/en/gen/topics/power_ps1q01-redhat.htm ), the archives of this and the khttpd mailing list, Ingo's answers to the slashdot crowd ( http://slashdot.org/articles/00/07/20/1440204.shtml ), and other pieces of information I could find. Is khttpd going out of the kernel and tux going in? How is the asynchronous (non blocking) i/o done? Can you describe the way the queue model works, I was a bit confused by Tiemann's article, but liked the way khttpd's model was described, ( http://www.fenrus.demon.nl/architecture.html ), but I am not sure if it applies directly to Tux. Regarding the two patents http://lwn.net/Articles/1251/ Atomic File lookup. I think I understand what the patch does, but why would you only want one file opened at a time? Is it to avoid blocking? Embedded Protocol Objects. I am confused. First I thought it just meant asking for a web page with all stuff included, get the dynamicly generated page and some static pictures back, but that's not how HTTP works. For each dynamic request, you send ONE "reply" back, but that can consist of both cached and dynamic parts? Example: someone requests http://server/tuxtimemodule, the module sends this back: <html> <head><title>the time</title> </head> <body> <!-- dynamic part begins --> 12:34:56 <!-- dynamic part ends --> </body> </html> Three parts, two very static parts and one pretty dynamic. Am I way off in my way of thinking here? :) I have more questions, but now I am very tired and must sleep. -- ___\ Jon Åslund