laarcnew | comments | discord | tags | ask | show | place | submitlogin

OP thinks Redis is an example of software that will remain at the current level of complexity over time. Can anyone think of other such candidates?

2 points by shawn 484 days ago


Editors: Vi, Emacs.

It would have been tempting to say Arc. But that's not true. Arc has always evolved with time.

In fact, popular programming languages tend to evolve rather than converge. C++, Rust, and Python are all changing in 2019. C seems like an exception.

Social sites tend to converge on a certain feature set with time. HN is constantly changing behind the scenes, but the core value proposition remains the same. (It's tempting to deviate from a core formula. But it's often a mistake:


I'm not sure. I'll throw out some recent ones I've seen. Fossil SCM tries to stay simple. All the separation kernels stay in around 3-10kloc range to enable high verification. QBE backend, a LLVM alternative, has that as a design goal. Might count Nanopass since they simplify the foundation with most of the rest optional based on cost/benefit. I was really impressed with lwan's claims [1]. The Lua VM probably counts. IIRC, Yogthos or someone else told me the Clojure language itself was simple with libraries doing most of the work. Any number of resource-constrained, embedded works [2] such as lwIP [3].


[2] I think you really missed your calling doing processor optimization and web apps. Your hobby interests are a perfect fit for embedded, software development. Might have even gotten you more skill and equipment for the hobby. Check out Jack Ganssle's The Embedded Muse if you don't already. I don't understand a lot of it but what I do is cool as hell. Mental comparisons with it will make you laugh when SW folks say stuff like unikernels aren't debuggable. :)



Welcome | Guidelines | Bookmarklet | Feature Requests | Source | Contact | Twitter | Lists

RSS (stories) | RSS (comments)