It's still super alpha, and I'm sure it will require a lot of bugfixes and future dev, but it's been really fun to work on some web software and move further in the indieweb direction.
@molly0xfff Just letting you know that while it is really cool what you've built (and alpha as you mentioned), context is lost on Mastodon when your post is split up into parts - the previous parts are not in the chain.
I had no idea what this post was about, and only got the context by visiting your profile and looked up the posts before this one.
Without context it becomes hard to follow what you're trying to convey.
The feed is built on top of the reading list software I've been running for a few years, and it ingests posts from Citation Needed.
I can also write posts in the microblog and automatically crosspost them to Twitter/Mastodon/Bluesky, while keeping the original post on my site. Like this (https://www.mollywhite.net/micro/entry/202403091817)!
@ntnsndr i actually do have platform-specific mentions! i built a little extension to the RTE i'm using where I can enter in a person's social handles — as well as personal website and/or Wikipedia page — and then the crossposter uses the appropriate handle depending on which network it's going to
Example of the mention tool on the sample text "This is a @mention!", with a dropdown with inputs for website, Wikipedia, Twitter, Mastodon, Bluesky, Tiktok, and Youtube
Screenshot of a Bluesky post by Molly White (@molly.wiki): ""There should be newspapers that we work at where we do this," says @katzonearth.bsky.social, who was the first to note that Senator Katie Britt had apparently brazenly lied while recounting an anecdote in her bizarre State of the Union rebuttal. flaminghydra.com/there-should..."
I just finally deployed something I've been working on for a few weeks now: a feed of my writing, posting, reading, and other various activity that lives on my website at https://www.mollywhite.net/feed!
I've admired the POSSE (https://indieweb.org/POSSE; "Publish [on your] Own Site, Syndicate Elsewhere") model for a long time, but only finally found the time and energy to put something together that I'm actually happy with.
anyone have any good suggestions for lightweight local testing when implementing ActivityPub? would be nice to test signature verification and simulate various activities without spinning up a whole local Mastodon server or prematurely deploying my super alpha software
@esk@molly0xfff Not as much my field (for my projects I'm deliberately only using the full, current RFC of HTTP signatures, for example), but I would point anyone at Jennifer's project and Helge's work, and I see that Jennifer already beat me to that:
@molly0xfff it’s truly inspiring to see what happens when an industry with multiple decades’ worth of broken consent models gets a terrifying amount of capital dangled in front of the hot new surveillance tech
one of the people behind this project is Max Howell of homebrew fame.
he also runs what seems to be some sort of AI-powered package manager, which generates incorrect and nonsensical descriptions for the packages it lists.
when people asked questions, they were met with an AI bot
dosubot bot commented 3 days ago • 🤖 Hey @pawamoy! Great to see you back here, diving into the intriguing world of pkg descriptions. Let's take a closer look together! Thank you for bringing this to our attention. I understand your confusion and concern about the project descriptions on pkgx.dev. Based on the information I have, the descriptions for the projects listed in the pantry repository are sourced from the package.yml files of each project. These YAML files contain metadata about the project, including its name, version, authors, description, and more. You can see an example of this in the twine package.yml file in the pantry repository. However, I wasn't able to find specific information on how these descriptions are generated or if there's an automatic process involved. It's possible that the descriptions you're seeing on pkgx.dev are placeholders or were generated in a way that doesn't accurately reflect the project's purpose. I agree with you that using the project descriptions from GitHub or the respective registries would be a more accurate and trustworthy source of information. This is definitely something that should be looked into further. I'm sorry I couldn't provide a more definitive answer. If you have more information about the process or if there's a specific part of the code you're curious about, please let me know. I'm here to help. Dosu Sources projects/github.com/pypa/twine/package.yml About Dosu This response is meant to be useful and
pawamoy commented 3 days ago • OK, the whole thing seems to be run by AI 😂 I'm actually impressed. This is a quite good answer to my question. Update: Only in the form, as it just repeats what I suggested, and the added info seems to be wrong. However I don't see any description in the metadata for the mentioned package, Twine. Going a bit further on my previous comments: the packages at the top of the page seem to have correct description. So I guess description are auto-generated while waiting for the system to fetch accurate metadata from various places or for human to actually add accurate description in the YAML files? It's weird because the description for platformdirs starts very well, and then: Platformdirs is a Python module that helps determine platform-specific directories, like a "user data dir". It offers functionality to automate workflows, host and manage packages, find and fix vulnerabilities, and provides instant development environments. It enables developers to write better code with AI, manage code changes, and collaborate outside of code. Platformdirs is available on GitHub under the repository platformdirs/platformdirs. Looks like the algorithm is: if there's only one sentence, fill the rest with auto-generated contents? Anyway, this is actually funny, happy to hear the thoughts of a human on this 😂 @dosubot good bot @dosubotDosuBot dosubot bot commented 3 days ago 🤖 Haha, I'm glad you're finding this interaction amusing!
daleglass commented 2 days ago Yup, this is clearly very wrong. Look at atuin, the first example. The repo is https://github.com/atuinsh/atuin It's not "virtual reality mapping software", but a commandline history search/archival tool. The description couldn't be less relevant to what the software actually does. @Ember Ember-ruby commented 2 days ago I get the generated pictures, waiting for maintainers to put their logo instead, but the descriptions... eh, something like GH default avatars would be better, as there's no chance of generating horrors, and doesn't need the resources that "ai" image generation does @alisonw alisonw commented 2 days ago Use Lorum Ipsum text if you just want to have content in a box, not AI crap which could be mistaken for real info. @odiferousmint odiferousmint commented 2 days ago • What is the point of dosubot? It agrees, and apologizes for not having provided a supposedly adequate response. it's always best to have human input for accurate and meaningful descriptions. Said the bot. sighs. @Delta456 Delta456 commented 2 days ago I wonder why no non-AI person is answering the questions. @leap0x7b leap0x7b commented 2 days ago Just based on the Atuin description, I can tell this is all AI-generated bullshit and not made by a real human being.
some maintainers have asked for projects to be removed. one didn't even care about the AI part, but was concerned that the package was misconfigured.
mxcl replied: "Presumably at some point someone will wonder why Rye isn’t packaged and try again at which point I'll point them to this thread and they can decide if they want to risk your wrath or not."
Remove rye #5360 Merged mxcl merged 1 commit into pkgxdev:main from mitsuhiko:remove-rye 2 days ago +0 −39 Conversation 2 Commits 1 Checks 7 Files changed 1 Conversation mitsuhiko Contributor mitsuhiko commented 2 days ago This removes rye (filed as rye-up.com) from the pantry. The package has the wrong name and also does not run the installer correctly. It is also not pointing to the actual canonical location on github either. @mitsuhiko Remove rye ad96c0f @dosubot dosubot bot added size:M request labels 2 days ago @mxcl mxcl merged commit 046e7d8 into pkgxdev:main 2 days ago 7 checks passed @mxcl mxcl mentioned this pull request 2 days ago I'm confused: what's with the project descriptions at https://pkgx.dev/pkgs/? #5358 Closed @mitsuhiko mitsuhiko deleted the remove-rye branch yesterday @mitsuhiko Contributor Author mitsuhiko commented yesterday Because you mentioned this PR in the thread I cannot reply to: I don’t mind that rye gets redistributed (when done properly). I do not know enough about this project to be able to fix this package here. This PR is unrelated to the AI descriptions which were not shown any more at the time of making this PR. @mxcl Member mxcl commented yesterday • The name defaults to our unique ID system which is domain based. We package many different ecosystems so name collisions are somewhat frequent. The name that we show can by changed with the display-name: key in the yaml/ We would always prefer that the build script by that the
Have you ever wanted to start editing #Wikipedia, but got overwhelmed or felt like you didn't know where to start? Every time I encourage people to start editing, I hear that, so I'm trying to help.
@molly0xfff
Done. One message left on a machine, one just kind of telling a staffer broadly why I’m against it (not much feedback). Both Dems. Hopefully it has some impact.
@mmasnick@hardindr@molly0xfff
I thought you agreed that the bounds of distributor liability were determined by the law (which I am advocating changing) and NOT the 1st Amendment. Are you know saying this is part of the constitution?
@DeanBaker13@hardindr@molly0xfff Distributor liability is still bound by the 1st Amendment, because it's liability based on speech. Which means that, per multiple 1st Amendment lawsuits, there are standards of what it takes to make a distributor liable. And you.. acknowledge literally none of them.
frustrating to see headlines about a "$290 million" hack out of even reputable security media like @BleepingComputer. actual loss is probably a tenth of that, at most.
@molly0xfff Accessing vendor specific device consoles through their web UI means I'm constantly closing that browser tab while hitting ctrl+w expecting to remove the previous word.