PHP UK Conference 2012
24 - 25 Feb 2012
at Business Design Center
Talk comments
Interesting insight into BBC's backend
Nice learning about new version, but it was basically a commercial about the new version without any actual SQL examples
Picked up some interesting things at this talk, how about making it a workshop next year?
Liked the talk, amazed at the hashing problem not being solved anytime soon.
How about putting out a reward for it.
Sponsors should be easily found IMHO
Like: $20K price for anybody that comes up with a lightning fast hashing algorithm that produces no collisions (or very few)
Well done for organising and directing the conference!
Loved your talk, consider organising a workshop next year.
I think most of us are really afraid of a situation that wordpress embraced: everyone commits to trunk, and can deploy at the push of a button!
However, maybe it is our own short-sightedness (or our manager's) that needs looking at.
After all, isn't that entire concept what Wikipedia is based on? Wisdom of the Crowd...
* Make it easy to undo changes
* Give as many people as possible access to make/undo changes
And the result:
* Critical mistakes are few and far between
* Minor mistakes get fixed almost organically
Definitely something I will consider!
I am sorry to do this to you James, but I was bored out of my skull during your talk. Might I suggest your title is slightly misleading.
Nevertheless being a public speaker doesn't come naturally to all of us, so good job for trying.
This talk could have been longer but really enjoyed what was covered.
Andy, the problem with that is the "very few" part. Even if a collision is super-rare, if it is predictable the bad guys can simply throw cpu power at finding the collisions. They only need to do that once and they now have a list of colliding keys they can throw at any server using that algorithm. So, it needs to not be predictable, but then we run into the problem I talked about related to the fact that we cache hashtables across requests and sometimes over server restarts. So it becomes a bit more involved. We would need to store the random seed with any stored hash table, which is of course possible and is likely the long-term solution to this. But it means a lot of pieces need to change so it will likely need to wait for a major release.