Well, that's one way to build a social network. Here's all the people you have talked to, communicated with online or been in proximity of. The problem with collecting this data is that once the government knows you have the data they will ask for it.
why didn't they just use Erlang? It's already proven and stable and it was designed to process reams and reams of data... not to mention it was made by a telco (I think) and it was open sourced in 1999
Yes, it's slower if you run it side by side with anything else on one machine, but I thought Erlang's main strength (in addition to reliability) was making it easier to create distributed applications on multiple machines (hence a better data cruncher)? Am I wrong? If I am, what was Oreilly making a big fuss about (and why do Hadoop/Java proponents like to mention Erlang)?
Well... it's a bit easier, but more than anything, it makes it easier in the sense of not screwing it up. It's not made for massive calculation applications though, like some of the more Fortran/C/C++ oriented things like MPI. It's oriented more towards fault tollerance - using multiple computers simply because they won't all fail / be offline in the same moment.
I don't understand why erlang can't do massive calculations when its optimized for distrbuted computing? I would think erlang would have a good messaging sys in place - it is optimized for dist computing afterall
yes in terms of individual machine benchmarks it is slower but it's built for concurrency and distributed computing (unlike the other languages you cite) - which means it can better handle do massive multi-computer computation/calculation jobs. (Parallel programming is hard, especially on languages not designed to make it easier from the start) O yeah anything that does distributed processing (including any libraries from other languages) needs to be reliable b/c if it isn't theoritically you'll either have to start your data crunching all over again or you'll have inaccurate results. If something is fast without reliability - what's the point?
Besides what do you think Erickson uses Erlang for? they use it for massive (and fast) calculations. C is very very difficult to make ultra-reliable, and Fortran is just archaic. which brings me to another point (the original one) if C and Fortran are so great why is AT&T not using it and instead designing another language? my original question afterall was why att is designing another language when they can just use erlang; not why other languages are better than erlang
