Supporting Unicode characters comes with a set of complications that you may want to address, as, for instance the infamous HTML-parsing with regex rant shows.

As soon as you start trying to address those, the simple problem suddenly grows quite a lot in scope.

Notice that HN already does support unicode: α β γ é ö 本

It actively filters out a small part of it, mostly emojis (which I think that it is a good thing to do).

I find this trend of despising emojis quite pretentious and annoying. They are very useful in some circumstances. Acting like anyone using them is a dumb teenager, or they will automatically ruin a community, is completely absurd.

Its Gen X and you know it

They don’t know it, but the rest of us do

Its not a trend its out of touch old people that dont know they’re out of touch old people yet

Is HN actually filtering these, or they use utf8 in mysql which is famously not full utf8mb4?

Hah, you think HN would use something mainstream like mysql?! I don't blame you.

But HN is 100% custom software, and built with it's own programming language as well, Arc (on Racket). AFAIK, HN still runs on files-as-a-db. You could check out the source here: http://arclanguage.org/install

That release is quite outdated. They added a lot of stuff to filter spam, voting ring detectors, and other moderation stuff. Anyway, as far as I know, everything is still saved to the file system.

Why "still" ? Is there anything wrong in using the filesystem as a database when it suits to do so ?

Nope, I'm not implying it's bad, just that they used to do that and they still do that to this day. I have nothing against storing data in files. In fact, I do that all the time myself too.

Don't databases also store data in files or am I missing the point?

Yes, ultimately they do (usually) and yes, you are missing the point.

"filesystem as a database" is not referring to database software that stores the data on disk but rather that the application is directly interacting with the filesystem. Imagine dumping a JSON document to disk, then reading it from disk, compared to storing a JSON string in a DB. Sure, they are both backed by the disk, but one is not "filesystem as a DB".

What exactly are we afraid will happen if someone uses an emoji instead of typing :) ?

it's not really a big deal. Just like not allowing bold or colored text in the comments.