Notices by ari :prideified: (ar@is-a.cat)

Today in silly matrix findings: you can have a user with only a single / in the localpart. which now means i now have a @/:is-a.cat user on my instance.

The requirements say that the only characters allowed are [a-z0-9._=/-] and that there must be one or more characters, but noone made a distinction for which characters the localpart can start with. The other characters would probably work as well :blobcatjoy:

I suppose the / was added to the list of allowed characters in localpart to make mapping usernames to kerberos realms easier :blobcatshrug:

here's the spec, if anyone's interested:
https://spec.matrix.org/v1.1/appendices/#user-identifiers

aperiodic reminder that you can instruct git to automagically use different url scheme, eg. to avoid having to deal with http(s) auth even by accident.

simply run git config --global --edit and add lines like these:

@icedquinn @49016 @sophie only reasonable if all of your contacts are on the same trusted mailserver. otherwise, email by itself leaves a lot of metadata.

@49016 @sophie sadly, there's kinda no good/clearly superior alternatives.
1) XMPP has too many e2ee methods, and no client supports all of the recent ones even;
2) MUCs in XMPPs require keeping a connection active, which makes it not-better than irc;
3) there are some IMs with good e2ee, like signal, but they require a phone number to use them, which is Not Great for some
4) discord is borderline spyware, and i'd expect things to get worse as they're nearing IPO/after it
5) IRC - to be usable at all - requires having an always-on computer somewhere, to act as a bouncer/proxy. And there's no e2ee standard that's actually adopted by a significant portion of clients.
6) some of the IMs requiring a phone number like to tout how private/secure they are, when in reality they're not

:blobcatshrug: the state of IMs is - generally - sad

@domi @linus :blobcatthink: i wonder how opposed to such an idea would konsole devs be.

@domi @multisn8 there are some bsd fanentities that don't like it when someone uses a more modern shell for scripting.

@q66 would having two usb sticks, one plugged into the keyboard, and the other into the laptop's usb port, help?

@puniko :blobcatthinkingeyes: Do these stats account for mastodon instances with emoji reactions patchset?

@puniko nyaa

> I'm in Jasień

Do you even realize how little it narrows things down? https://pl.m.wikipedia.org/wiki/Jasie%C5%84

@puniko nyaaa :neocat_boop_cute:

@puniko mrrrrp :blobcateyes:

@lanodan @Ollivdb meanwhile ruby:

but here's the kicker: it's allowed by ieee-754

or, to be more specific: you can trap on operations which produce problematic results, but you don't have to.

Just python being sus:

@puniko Nyaaa! :blobCat_tilt:

@puniko mrrrrp :neocat_boop:

@puniko not sure why but you aren't the only one.

both this

and this:

is valid (doesn't throw obvious errors) python code.
tbqh, I would've expected k, v to be out of scope in the second example, and the function to throw an error.

@q66 yup, and this one was especially annoying as it was an artifact of moving code around (-> changing indentation) *and* still returned seemingly valid data

average fedi user looking for a new distro

pic cw: anime