Re:C/C++ (Score 1)

+1. This is a very good point. I'm shocked with how hopelessly out of touch most of the above comments are with reality. If I need to write a Lattice QCD simulation which will take weeks to run and consume gigabytes of memory, I would accept no other language than C. That being said, desktop applications are not HPC applications. Far from it, the programmer's time frequently matters far more in desktop programming. While higher level languages like OCaml, Python, Ruby, C#, Haskell, and the rest may be slower, from a desktop perspective this should hardly be noticeable in a properly designed desktop application. If the user is seeing any type of latency whatsoever, whether from IO or computation, you are doing it wrong. In particular, I think the GNOME project's dependence on C for so much of its stack is a tragedy. The widespread usage of such a low-level language like C for desktop applications has the potential to open up security holes (unchecked buffers anyone?), crashes (anyone use Evolution?), horribly structured code (anyone look at some of the older gnome platform code?), and just plain stagnation (it's not easy to maintain such a vast codebase of C). Moreover, gobject itself is most certainly not as well optimized as a compiler could do. The lack of compile-time information requires it to be be extremely generalized. While this is in the name of ease of binding, even most bindings don't map to gobject in a reasonable way. If I am programming in python, I want my code to look like python, not some gobject/python hybrid (although the python bindings are admittedly improving, albeit slowly). I think that the original poster brought up a very informative point. I think what is even more interesting, however, is how quickly the /. crowd began to raise strawmen to attack. Trying to have a debate about proper languages for desktop programming when people are set on developing for HPC is impossible. This idiocy is what makes me dread reading slashdot comments.