Guys … platform support is a really big issue for us. And …
We're totally committed to making sure that everything we release is of the highest possible quality. Each additional platform we port to…and this means Linux platforms too...increases our QA, docs and release engineering burden. It's possible to get things to compile on some pretty remote platforms. Look upthread and you'll see we have compiled on platforms like MacOS and Windows (although this does require the application of some pretty heavy duty beatings). The problem is it would take us (say) about 3 months of effort to get the QA, docs and whatnot up to the standard we expect. And meanwhile, our efforts to get new features tested, and the reliability / robustness engineering we need to do, would suffer.
We're making a deliberate, mindful, strategic decision here. We're prioritizing features, quality, and enterprise readiness over availability on a broad range of platforms. As we mature, of course, we'll add more platforms.
Also … what Bryan said. Although the SciDB "core" runs on Ubuntu / CentOS, there's no reason you can't deploy your 'R' client or your Python code or your Java-via-JDBC clients on Windows.