Notices by Glyph (glyph@mastodon.social)

The Atwood Dial Principle continues to hold up https://www.heise.de/en/news/Mastodon-Founder-Steps-Down-as-CEO-Receives-One-Million-Euros-11081969.html

I need an invoicing solution.

I want something that can send emails, schedule reminders, nicely format PDFs, and has a well-documented API for uploading line items from time tracking.

I've used Stripe invoicing for this in the past, but in the situation where one has clients that use *other* payment mechanisms, IME this creates annoying reconciliation gaps.

Any recommendations?

- upgrade Emacs
- launch it for the first time
- really slow, laptop getting warm
- "oh, it's a system update, it must be Spotlight indexing"
- realize that that thought is insane
- look at CPU graph
- yup, it's Emacs???
- switch-to-buffer
- "*Async-native-compile-log*"
- oh, I guess this is just … everything, now

@dalias @miss_rodent yikes that is a very unfortunate coincidence

this is not me saying that the VCs are particularly good at predicting outcomes. this is a "house always wins" situation.

a) it doesn't work, but you get dependent on it for achieving the same result; your skills atrophy, they jack up the price, they make a ton of money, they win

b) it works TOO well; you actually get way out ahead of proprietary alternatives and start harming their bottom line. they jack up the price so much that only proprietary alternatives can afford it, they win

why do you think those venture capitalists are subsidizing this thing. do you think that they are hoping that you will just increase the collective social value of the commons, and rejoice? https://mastodon.social/@juliank/116974045503028965

I tend to believe that "superintelligence" is a meaningless term and "AGI" is functionally unachievable (certainly with the current generation of technology), but nothing makes me believe in the superintelligent machine-god than Them sending Their time-traveling agents to tell us all, wild-eyed and panicked, that we have a "moral imperative" to use a specific product to achieve a specific goal

the statements "LLMs are only good for producing tedious and borderline-worthless bullshit" and "LLMs can produce code of vaguely equivalent quality to the current software industry median" are not necessarily mutually exclusive

I used the word "fresh" here just to mean "novel" as in "written more recently than the aipocalypse" but the more I look at it the more I am realizing that this is maybe a great word to use? There are two kinds of information, fresh and slop? Slop code is always somebody else's reheated nachos, even if you don't know where the nachos originally came from, whereas fresh code is original.

they earnestly believe we’d be happier if we did it too, and they want us to enjoy the sweet relief of letting our minds go along with them. this is not to say that it’s harmless, you still shouldn’t do heroin even if your buddy *sincerely* enjoys heroin, but I think it can inform our resistance

so the mustache twirling LLM capitalist would be overjoyed to “respect the decision” of the codeberg crowd, make lots of noises about everyone using the tools that work best for them, and quietly continue hoovering up fresh data. instead they are high on their own supply, telling us to try some

RE: https://mastodon.social/@404mediaco/116986805020926545

this is one reason I ascribe more incompetence than malice to most boosters. the boosters were mad about the codeberg ban. a calculatingly evil conspiracy would understand that a brutal model collapse is inevitable for coding models if there is no fresh, nonslop option for modern code.

@ecadre @DanielMReck GNU emacs was a relative latecomer to the party, and not the first Emacs written in Lisp — no matter how you slice it (rms, the name “emacs”, a lisp editor) developments were happening in the mid 70s: https://emacs.brause.cc/timeline.html

@thomasfuchs what are you doing for input device verisimilitude?

one day somebody is going to invent a database where schema migration is not a full time shitshow of a job for three whole people, and I am going to make so many cool apps

OK folks, I need some help from the #Python #OpenSource community.

I need a code review on this: https://github.com/twisted/klein/pull/594

You do not need to have any special qualifications other than a general interest in web frameworks. Experience with Twisted would be helpful but not required.

Some background… 🧵

This is a MASSIVE, years-old PR that adds a ton of documentation for how you can actually USE the framework, including adding enough of a built-in database adapter for the basics of session and user management that you can see a working example without intuiting a bunch of interface implementations from scratch.

The Twisted team doesn't have a ton of interest in the user-facing bits of our "microframework", Klein, which is a sort of Flask-alike for Twisted. Flask with Twisted characteristics, if you will.

There isn't a ton of interest from outside the community either, as the framework is a bit of an incomplete shell, having *parts* of form-handling and session-management but requiring you to build the other parts yourself from a barely-documented construction kit.

Also hey @davidism while I have _zero_ expectation that you would actually want to do the review part of this yourself, I thought you might like to know that That One Other Thing That Uses Werkzeug might actually become more generally viable

Today, I've unvendored the half-finished version of https://dbxs.readthedocs.io/en/latest/ and made it a dependency.

You are welcome to review other bits of the code as they interest you, but the main "approve" stamp I'm looking for before I land & release this is someone to read the documentation and say "yes, this basically makes sense". Extra bonus points if you actually follow the tutorial.