Downfall scenario. Never trust an infographic you haven't faked yourself - that's probably what the manager magazine thought when presenting the "evidence" that Apple is investing less in research and thus jeopardizing its future viability - while R&D expenses in Cupertino have increased, but in the representation in relation to the even more increased revenue, it then looks appropriately inappropriate. Only blogged here because probably again a lot of people will feel the need to tell me with the graphic why Apple would be bad.
Archive 30. June 2011
Opa: Advancing web development to the next generation. I think I mentioned this in the old blog, but what the heck - now the code is on github and under an open source license. And the entire project is not just something that has recently come out - it has 10 years of research behind it, the people really know what they are doing. The idea is cool (and has already been adopted by others as well): a language from which all layers of web applications are generated, including JavaScript, HTML, CSS, and the backend. All of this with type safety and corresponding checks and, for example, resulting security against injections and XSS attacks. And the language runs on OCaml underneath, which is also not the worst language. I should really seriously deal with this, especially since distributed installation comes along with it - and thus a far easier scalability is given than with many other approaches (the one from OPA reminds in a certain way of the one from Erlang-based systems). Oh, and web applications are then simply just a single executable - and not gigantic directory structures with thirty-nine XML files (or YAML files) to customize.
SymPy. Symbolic mathematics with Python directly in the Python shell. Together with NumPy for matrix operations and SciPy for various other scientific functions, Python becomes quite a respectable mathematics system. Nothing really new except perhaps SymPy itself.
PyPy Status Blog: Global Interpreter Lock, or how to kill it. Wow, that sounds good - the PyPy project wants to mitigate the GIL through STM (Software Transactional Memory) - and has quite an interesting approach for that. Some things are only possible through PyPy, because the corresponding infrastructure for code transformations is available there, which is missing in CPython.