iansjack wrote:
I get the impression that your comments were initially predicated on the number of trivial questions being raised of late. In reality, most of those questions could have been answered by consulting the existing Wiki in the first place, so I don't think they are result of any failings in the current status quo.
It's true, that's the origin of my comments. If you/I see the questions from ~2007, you'll find more interesting ones, with more diversity, more theorical stuff, etc etc... And it's true that
most of those answers are easily answered by the existing wiki. But topics related to I/O or Filesystems or Network (I would say about 20-30% of new topics) are simply lacking in the wiki, or pretty wrong and inaccurate (
Select your cup of tea here)
iansjack wrote:
I agree that the Wiki doesn't provide a good in-depth description of all topics, but I'm not convinced that it should. It certainly provides enough information to get people thinking and to give them ideas as to what they should be Googling for. I don't think a "Painting by Numbers" Wiki would be a useful contribution. People have lots of different ideas about OS development and they should perhaps be left to develop thos ideas rather than being given the step-by-step instructions that some demand.
Sorry if I gave a wrong impression. I'm not intending to have a "step-by-step Wiki". That's just a stupid idea. Let's put the TCP/IP stack example. In what I think would be a better wiki, there would be a Internet Protocol page (was writting that yesterday, I'll upload it later). That page would
summarize the IP header (pretty simple...) and some comments on what each member does, including pitfalls. Details (not important ones) would be redirected to the specification. Now let's put another example, the Transfer Control Protocol (TCP). The geniuses who designed the stack back in 1981 decided that TCP would do all the hard work, and it's true, it's a pretty more complxe protocol. Its wiki page (as far as I know, no one is writting it) would contain essential and summarized information, i.e.: what is its purpose, basic port concept, the holy header, etc etc... But always refering to the RFC as a first resource for
completeness. That's the idea, having completeness through references. There shall
not be any information outside of what's simply true and what the protocol mandates. No "step-by-step" or "tutorial" stuff.
iansjack wrote:
But I take your point that it is sparse in subjects such as TCP/IP. You seem to know a lot about that so why not put your money where your mouth is.? As for numbers, I'd guess that most forum membership follows the iceberg pattern. And, again, most forums have a number of users who just ask questions and another set who just answer questions. That's just the way it is.
Completely true.
_________________
Happy New Code!
Hello World in Brainfuck :
Code:
++++++++[>++++[>++>+++>+++>+<<<<-]>+>+>->>+[<]<-]>>.>---.+++++++..+++.>>.<-.<.+++.------.--------.>>+.>++.