I wish more web services provided opensource versions (Service-as-a-Software). Large orgs would still pay for service contracts even if they self host, and just having the option makes everyone more likely to invest on top of your service.
Chat at "indie scale" is fairly easy. 10-20k concurrent users fit into one or two machines with off the shelf software (ejabberd) with minimal expertise. 1k-2k concurrent users could easily be done in almost any stack.
If you're doing more than 20k concurrent users, odds are you can afford to hire someone to scale up or build out a team to do it, or find a 3rd party service and pay them to host your chat.
The author works for mozilla, and firefox is a common theme on his blog. I think it's for regular readers of his blog to understand the practical impact (readers he assumes care more than average about firefox).
Found this after Joël tweeted the quote "The dilemma every artist confronts, again and again, is when to stick with familair tools and materials, and when to reach out and embrace those that offer new possibilities." So true.
If there's one thing I've learn from reading about Systems Theory / reading Russell Ackoff, it's that every field, every practice, every art is far more similar than they are different.
"This method takes a foo object, applies HTML encoding, and returns an array of the original and the clean code." could be a 3-line unit test that's just as clear to read, but has the side effect of actually being true instead of maybe being true!
I would consider that comment just as bad as the ones in my examples, but maybe not quite as obviously so.
Actually, that's EXACTLY the case I had in mind. Obviously I couldn't put a 50 line function with weird data structures in my blog post, or it would be unreadable. But the point is that comments for that function are bad! They'll rot if you ever change the function or the assumptions at all. Instead, you should break the 50-line function up into smaller functions, and add assertions and test cases for all of those "magic-lik assumptions." Then get in the habit of reading tests first. Now your code-as-comments will never rot.
Thanks for the clarification. The downsides to that approach are that you need multiple machines, and the duration of your deploys is much longer. Not to mention, you'd have to script a deploy process across multiple machines (which is not easy, in the way that "SIGHUP Gunicorn" is easy).
Personally I've found the "put new instances into a load balancer" method to make more sense for system changes (packages, kernels, OS versions) where deploying the change is inherently slow or expensive, but the method doesn't make sense for code deploys where deploy time is important.
Your method will stall responses for server shutdown + server startup time, which for Ruby/Python apps is usually measured in tens of seconds, and for other web servers can be much worse. Hot code reloading lets you avoid any downtime at all, and with it usually being built into the framework/language specific server you get the functionality "for free".
Zdd (the project I linked to) is all about spawning a new process in parallel. All the advantages of your approach (switch to an entirely different language? Who cares) but without the stalls.
Zdd also lets you keep the old process alive through the duration of the deploy (and after), and with a little work could let you switch back in the event of a bad deploy without having to start the old version up again.
By far the easiest way to get similar behavior is to run your Django app via Gunicorn (proxied from Nginx). Gunicorn supports hot code reloading via SIGHUP, and it does so by forking and gracefully killing old processes.
If your requirements don't match gunicorn (not django, not python, etc) then you can use https://github.com/TimothyFitz/zdd a project I wrote to automate rewriting nginx config files to deal with changing proxied portfiles. To integrate any existing server, all you have to do is make it bind to port 0 (let the OS choose a port) and the write a foo.port file that contains the port number (like a pidfile). That's it.
One really impressive thing to me about 4chan's infrastructure: All traffic is served via 4 machines. Two webservers, a database and an admin/cron box. 22M uniques on 4 machines is insane. Hopefully Chris will write more about his setup some day. (Note: these boxes were directly serving all traffic up until ~6 months ago when Cloudflare was introduced)
Do we need significant software changes to take advantage of the new power? Are the TFLOPs somehow not directly comparable?