#turtle.io

This major release of turtle.io features breaking changes to some core APIs, to correct the oddness in the data flow that crept in over time as features were added. What was once a series of synchronous functions which occasionally branched into streams, is now an optimized asynchronous pipeline that utilizes Promises to effectively schedule the functional units, so the hamster wheel will spin faster for everyone. The initial results appear to be a marginal increase in concurrency; turtle.io was already hitting >99% ... at least I didn't make it slower! yay!! I've also reorganized the multiple class source files into a new single "turtleio.js" file, mostly for IDE completion.

If that title piqued your interest, the answer is yes! turtle.io can now consistently hit >99% in most cases. That's more than twice what you'll get from express.js, or nginx, or pretty much anything else that's remotely popular at the moment. There are faster servers on the JVM, but they're less durable and what I consider true edge cases (limited value in the real world). So what does that concurrency get you? Peace of mind really. Your code should be CPU bound, and turtle.io ensures that it won't "get in the way". If you're serving static files from disk, or if you're writing dynamic apps and don't want to create your own error handling, turtle.io is a serious contender for your consideration. tenso is a project built on top of turtle.io which makes API facade creation super simple, check it out if you want to see what you can do with turtle.io. turtle.io

I revised my benchmark method for these servers (disabled stdout), and switched to a slightly weaker i7 CPU. The results show that only vert.x is in a position to beat turtle.io, but it's inconsistent availability caused the test to fail multiple times, before I did a final run to collect results in a single screen (only cropped out socket errors). https://www.flickr.com/photos/deviation/15396125973/ Transaction: turtle.io - 1,498.38/s nginx - 1,115.62/s express - 1,144.67/s vert.x - 1,640.76/s (failed to complete) Concurrency: turtle.io - 99.25 nginx - 26.58 express - 42.32 vert.x - 65.62 Overall I think this type of test represents real world usage, and turtle.io is the clear winner at this hardware spec. This also shows how different node.js modules can impact performance, as turtle.io & express simply wrap 'http'.

This minor patch release adds 4-5 HTTP headers to the average non-304 response; two are security focused, one is for Internet Explorer (sigh), & two are standard headers that a developer may want for context. Iframe embedding is blocked by default; it can be re-enabled with your server configuration. For the end user, browsing the web, there is practically no difference. turtle.io

Last week I released turtle.io 2.0.0, which added supported for Connect middleware. Out of the box, it's now extremely easy to host many static websites, as well as mix middleware on different routes, for different virtual hosts. If the middleware doesn't terminate the connection, it'll fall into either `respond()` or `error()`. turtle.io

I'm glad to get this release out! It represents a shift away from my old framework "abaaso", toward my new one "keigai". There was no new features added, just bug fixes and a motor swap. turtle.io

turtle.io 1.1.0 was released the other day, and it addressed regressions in the sessions, which unfortunately made it into the 1.0.0 release. The `cipher()` method was deprecated, and has been removed. Today, 1.1.2 was released, which fixes a few minor error edge cases, which was found thanks to StrongLoop. turtle.io

I'm happy to write that turtle.io has matured into a stable 1.0.0 release! You can simplify your node.js stack with turtle.io by letting it be your web server (with virtual hosts), cache server, & reverse proxy. turtle.io