Very interesting... and pretty nice to see the work in supporting multiple storage engines... In particular the levelup/down implementations are interesting as there are interfaces supporting multiple backends.
It makes me curious what it would take to support a REALLY thin server-side solution for only what is needed by leveldown's API over a TLS connection, optionally with replication/consistency.
Not sure if this would just wind up being another etcd.
Interesting... I hadn't thought of running an application in dev via a mount point. With a base image that doesn't change much, this could be very useful.
No mention of reduce...
Object.keys(obj)
.map(k => foo(k, obj[k]))
.reduce((a, b) => Object.assign(a, b), {});
that's one of my favorites... break up an object/hash map, do something to each, and reconstruct hash map... assuming `foo(k, v)` returns `{ [k]: bar(v) }`;
map is cool and powerful, but so is reduce.
I know... I work with Windows, Linux and Mac on a daily basis. I was just pointing it out in context as to why that was chosen. It *could* also be thought of as the *home* for the application itself.
I even mentioned that node specifically doesn't do anything with tilde in the fs or path modules, even in a unixy environment.
I'm relatively sure the author has exposure to .Net as the tilde is used as project root the too. It doesn't work out autosub for profile/home in node currently in any of the path or FS modules.
TFA's example doesn't work for server-side fetching, and kind of dismisses the issue... the reality is that one should use methods that work for both, if you're going to have universal rendering, so you don't have weird complicated code to do the same thing in two places.
No mention of reduce... Object.keys(obj) .map(k => foo(k, obj[k])) .reduce((a, b) => Object.assign(a, b), {}); that's one of my favorites... break up an object/hash map, do something to each, and reconstruct hash map... assuming `foo(k, v)` returns `{ [k]: bar(v) }`; map is cool and powerful, but so is reduce.