New Tools for Old Problems · These are questions for wise men with skinny arms

New Tools for Old Problems

Two major thoughts here:

Have you seen the Azure ML workspace? It makes doing machine learning easier than setting up Matlab. There are other machine learning services hosted by cloud providers, but this one can go from no coding required by dragging boxes and drawing lines all the way to an integrated environment for Python and R. This just blows my mind. I thought that Matlab was going to be the easiest way to set these sorts of things up, but this just makes it stupid simple. It’s not a platform for research, but it has very practical tools to perform the machine learning process. And it’s made by Microsoft! Who counted them out? It’s a freely available platform that uses open source tools and cutting edge machine learning techniques. I checked up on other projects of theirs, and while I don’t think the Hololens is going anywhere in the next decade it’s good to see that they are moving in a positive direction. I’m no MS fanboy, but I’m glad that they are hanging in the game just to see if the other other giants lose their halo. I’ll make a tech babble post later on some of their tools and why they are worth using.

Rust hate is going around. Rust has been at the bottom of my list of languages to try after I saw it in the alpha stages. I did the same with D because of my experience with C++ has made me look for something profoundly different. I’ve been seeing Rust pop up again and again touted as a ‘true systems language’ or something like that. It gets compared to Go and Nim and the other hot new languages and I’m still not sure if people are serious about it or are just excited to try out a new toy. I’m not sure what the purpose of a safer language is if it doesn’t offer serious usability or tooling advancements over existing safe languages. Rust just doesn’t have a compelling feature story, much less a success story. Memory ownership? Really? As something the programmer has to manipulate and consider just for the program to compile? If that offered a serious performance boost or some sort of great interoperability because it made tooling easier than maybe I’d be willing to take on that cognitive burden. But saying that we need a new language that is memory safe but doesn’t use a garbage collector just looks like it’s trying to fill a imaginary gap left by existing languages. I’d compare it to Arc vying to become the ‘best’ lisp because the others were so fragmented. I’ve read that there were a few innovative ideas that were scrapped to make sure Rust shipped in a reasonable time-frame with reasonable benchmarks. Looking at it now it should have baked a bit longer to grow something really unique as the ‘core feature’ just isn’t innovative enough to attract the kind of attention a language needs.