There seem to be lots of people here that are offended by the "Node.js cancer" line. I am not defending his choice of words, but in TFA he wrote "Unix can do that all by itself with out all the stupid Node.js cancer.". I take it as, "there is no reason to have node.js involved since simpler tools can accomplish the same task". This says nothing about Node.js itself. In fact, try re-reading that same sentence but substituting a different technology from the following list: [MySQL, MS SQL, PHP, Perl, Visual Basic]. Does it sound slightly more reasonable now?

From what I gather his gripe was with the fact that the original Hubot was over-engineered. I agree that it was. Not only that, but the original Hubot came with almost no useful documentation, and a pretty horrible bootstraping mechanism (scripts that download other scripts onto a server over HTTP should never be run in prod environments). Getting it to run one step away from how it was intended (specifically on IRC) required digging deep into the code since the command line options/environment variables that were supposed to be passed in to make it tick were not exposed in any docs. This daemon process did not background itself properly, writing a PID file, etc. The whole thing was a cool proof of concept, but running it in production would make me very nervous. On top of that, it did require Redis, for some ungodly reason. SQLite would have been a much nicer choice since it's much more easily obtained (at least on CentOS) and does not require a separate process or having to configure. (Offtopic: has anyone had good experiences with SQLite style no-sql data stores? Something like MongoDB, but storing stuff on disk and accessing it in-process?).

Don't get me wrong. Hubot is awesome. It's just that it's only awesome at GitHub, where it integrates into all sorts of ecosystem and the culture. I tried it for an afternoon, and by default it comes with no useful plugins (mustache generators notwithstanding) because the plugins are the trivial part when you have a great infrastructure already. Imagine if I handed you the source to bash, but you and everyone you know uses MS DOS. Yes, you can see that bash has potential to be powerful, but until you have utilities like find, ls, grep, awk, sed, ed, cat, and so on.