For French readers: I have created a HN like for French news with the same interface than hacker news and a similar time/point based algorithm for auto sorting most popular articles.
It’s not mobile friendly although maybe this says more about mobile browsers than the site itself, because it looks like a typical old school “mofo” style website (h1 h2 p etc. no css)
Running a local benchmark to compare Node.js and Deno gave me the same magnitude of performance difference. I like the concepts behind Deno but the performance should stay a top priority. Even more for a new technology that is looking for future adoption. If Deno gets faster than Node.js, I adopt it. If it stays 5x less performant than Node.js, I skip it.
Are you running deno 0.10.0 versus Node? Since there has been some internal refactoring it should now be 80% wrt basic http req/sec (ref: https://deno.land/benchmarks.html#all , though the benchmark might have not covered everything)
Overhead from Flatbuffers is a major reason of the slowdown and we are seeking to get rid of it.
That should be an ideal application. Usually the trouble with 3rd-party garbage collection is that it has to discern pointer vs. any other machine word. That's more of a problem with C; this is why the Boehm GC library calls itself "conservative". A runtime like V8 can follow a spec when it allocates memory so everything is properly marked.
You might not even need magic lantern for that. I have a 77d and it streams a live view to the Canon EOS utility over WiFi. Quality seems pretty good. I haven't looked into it myself, but I wouldn't be too surprised if someone has written software to stream from it in this mode.