1

I am currently thinking about what would be the best way to get people involved into exchanging knowledge about optimization techniques (for me personally in C++, but it would probably be just as useful for other languages too). So before I come to the core of my question, let me first tell what made me think about it.

On support sites and/or in irc, I often find people asking for help in optimizing a certain piece of code. Sometimes this is offtopic, or smells too much of "gieve teh codez", sometimes it is just premature optimization. But all seem to have in common that people have some code (ideally selfcontained) that someone could modify, and post back. And all of them have people that don't know much about optimization techniques, or worse, are full of optimization myths and wonder why when applying them nothing happens.

Then there seem to be quite some people that know about optimization techniques, and some even like to apply them, but don't have enough opportunities to do so. You can see some of them at sites like euler project where they want to not only correctly solve a given problem, but in a fastest way.

So my idea was: get those people together, let them discuss, and spread knowledge. Let them formulate real knowledge, and extinguish myth. And there is a lot of myth to bust.

So the core of the question is: Is an SE site, with its Q&A format the right tool? Someone I talked to said it would be awesome if people could try out their code like on ideone.com, but actually with the snippet in question being accurately timed, so that people would get their reputation also from the actual speed they achieved. But that would cost hardware, probably a lot when it gets attractive, so besides the question who would pay for it, there is the question whether this would just a gimmick to play with, or something really necessary.

Or in other words, would the SE format already suffice, not only to get questions answered, but to form a solid base of knowledge about optimization techniques that people can reuse?

1

For the most part, you could simply ask your questions in the tag on StackOverflow (link).
If the original StackOverflow does not help to build up the knowledge base that you desire, than I doubt that another Q&A site with the same engine would improve that situation. So the answer to your question becomes "No".

For busting the myths around optimization, asking for and providing hard measurements is probably the best bet, at least until SO implements a system like ideone.com - which I don't expect anywhere in the near future.

I like the idea of being able to try one's code in the browser and gain rep directly from the measurements.... but I think it's very prone to abuse.
I guess CodeGolf comes closest to what you have in mind, but it is still based on the upvote/downvote system.

  • The problem I have with posting those questions to SO is the relatively low SNR. A lot of questions receive "myth" answers, and there is very little discussion and "knowledge building". codegolf.stackexchange.com/questions/3950/… is also one of the examples of answers I do not want to have. What I have in mind is mostly aiming to spread knowledge about optimization and make myths disappear. No more "use ++i instead of i++" but more "use this O(1) algorithm instead of your O(nn)". People on SO also often start focussing on non-issues. – PlasmaHH Nov 22 '11 at 13:57
  • 1
    For finding an algorithm with a better complexity, how about Code Review ? – S.L. Barth Nov 22 '11 at 14:11
  • Updated my answer in the light of your comment. – S.L. Barth Nov 22 '11 at 14:28
  • I will monitor Code Review a bit, it seems not like the perfect solution, but might work for some cases. While it seems to produce satisfying results for the peoples actual question, it is missing the "knowledge building" and "myth debunking" part a bit, and also the discussion about optimization techniques. The later one possibly even being offtopic there. My "great wish" here might also be that people all around know more about actual optimization, not about useless myths. – PlasmaHH Nov 22 '11 at 14:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .