Working on DayJob code, I noticed how one file is like nearly 3,000 lines. Now for my own projects, I have this "thing" where I'm reaching nearly a thousand lines of code (including comments) that I feel compelled to move it into another file. That's if I notice it. I barely do nowadays but when I do, I'm always a bit shocked—for no particular reason outside “wow, that's a lot of lines”.

I do think hosted databases are going to die out over the next 5 – 7 years for B2C applications and move to allow people to store data on their devices and just collect metrics. It's cheaper for companies. Biggest issue would be handling “migrations” but eh.
I'm combing down the list of open issues on the Microsub tracker, and I was thinking again about how I'd want to “simplify” the experience of people following one another. I thought of this mainly for my implementation of Microsub, but I think that having the ability to preview who to add and being able to make a 'meta-subscription' can help. This would do the work of discovering the feeds at a URL, subscribing to all of them but showing them as multiple ones in a reader. The reader would need to know the original feed to recommend its whole removal, which is something that's tricky.

Something I think that's slightly forgotten about DNS is that yes, we have a centralized form of it that we use every day. We can also encourage decentralized forms of it (without web3, dear God) in conjunction to the federated ones we use every day. Granted, companies tend to use this for internal development environments and I use nss to give my local Docker containers resolvable host names. If we had a way to specify what resolvers for DNS we could use in a computer as simply as we can set up WiFi, we could see a way for people to encourage “private” DNS with a one-click setup. Obviously, a good desktop environment would allow for an option to 'reset' this in the event the upstream resolver fails.

Made excellent progress on this. I'm becoming more aware of the need to have JF2 support to support other places that things can be piped from. It'll also become more required as I begin working on the Microsub facets of Lighthouse. I've managed to not focus on it, just it. I can also see it simplifying the logic for my templates on my site if I use that, but I'll have to see.

• posted archived copycurrent

I do think that I might entertain the idea of using @context in my MF2+JSON when rendering my feeds. It's a small change that promotes MF2 in the JSON form and it makes it discoverable when attempting to parse (outside of content type hints).

How many of you all be using Twitter Blue? Seeing Facebook's Instagram attempt to launch paid subscriptions makes me wonder how viable those things are (and if people are seeing actual numbers)?

by • posted archived copycurrent

I ask because after seeing, that sort-of completes the trifecta of the big social media platforms integrating paid subscriptions (versus free ones). Things like KoFi, Liberapay et al work just as well but require people to have alternative accounts.

This is something one probably learns in OS fundamentals, but there's like only three file operations that exist: reading, writing and deleting. Everything else is like a combination of these.

I mean, even deletion could be a special form of writing, but I don't know what that'd look like.

Was thinking about how to fetch Webmentions and I think that this approach is not only fantastic (obviously biased, lol) but now once more implementations pick this up, static sites can provide a feed of their Webmentions with one line! That's truly my hope—to be able to query Webmentions (and subscribe to them) so people can see reactions to things in their social reader. An intelligent reader could filter out some of those replies (from undesired sources) and still give a relatively healthy experience.

You know, I understand why we use UIDs for foreign key identifiers. I understand if you got like a billion rows or if you never actually delete anything. But like if you know it's small, don't bother.

I'm glad that SQLite nudges away from concrete types and uses storage types. Because the need to test values is pushed up into the application layer, and I don't think I like doing a lot of strict typing in the database (though I've been bitten with and without it)


This is something I need to come back to during lunch today.

by someone • archived copycurrent

Okay, checking this out now. Overall, this is a superb piece about language. I'm still reading Because Internet by Gretchen McCulloch and this reminds me a lot of her work.

Some things I wish it highlighted:

  • The use of language to hide access from those who don't grok things when it comes to the Web (or just media—like there's a difference between manslaughter and murder and I only found out a few days ago)!
  • I love this bit — “There is so much implied in how we write, and plain language should aim to make the implicit more explicit.” This is something I need to work on myself, and I wonder if it's an attribute at being good at either communicating or writing.
  • Black-box solutions meant to work for the masses can't do their job. The masses need the ability to review, adapt and configure said solutions. Things like that is why the Lexile Framework for Reading is trash from that short demo.

The Pudding is also a very interesting site! Definitely subscribing to see more from them.

“Onboarding” is one of those things that I wonder is a collective psychosis that we've put on people to think there's a “right” way to do something. Like sure, there's very wrong ways, but how can something so specific be generalized so much? I'd love to learn more.