And just to be very explicit why this is an issue: each time the package is upgraded through an automated update, the PKGBUILD may change (e.g. to adapt to different dependencies, file structure, etc introduced with new app version).

That also means an AUR maintainer can smuggle in malware with any of those updates, even if you checked the original PKGBUiLD when you installed. And, anyone can request taking over maintenance for unmaintained packages, so it can even happen if the original maintainer was benevolent.

Always check PKGBUILD files on upgrade, even if just a glance. If I remember correctly yay had a function to always show you PKGBUILD diffs before updates, not sure if that was automatically enabled.

isn’t the better solution not to accept PRs from unknown / untrusted sources

I think that’s partly the point of this exercise - if they find a meow they now know this is an untrusted source.

Because it’s pretty easy to say ‘ignore untrusted sources’ but when you’re maintaining an open source repo (especially if it’s still pretty small/new) this detection is part of the cognitive burden. Almost every contribution will technically be from an unknown source for a long time, until, if you’re lucky, some drive-by contributors turn regular.

While the above is a little more focused on front-end/web projects I think one thing to keep in mind generally is also to meet the user where they are, or rather where they expect. E.g. if you create a tool/plugin for vim, users will be thankful for having a vimdoc for reference. If you create a command line tool, a man page will be helpful for quick lookups.

I know that many projects today (especially when first getting set up) focus on a presentable readme but in many cases there are tools to automatically adapt this markdown text to the generally preferred documentation style and location.

Don’t have to focus on perfection first but just having a pipeline set up which can then be iterated on over time.

For a more advanced approach I think you can get a lot out of understanding the principles of the Diataxis framework - but this is also larger up-front effort.

Thanks, that’s good info and makes a lot of sense.

Which certificates were they generally after instead?

Perfect, looking good!

There’s a nice Calvin version you can use instead:

Or this I suppose

Hold on, I’m a little confused. Are you talking about paperless the document management software (and it’s ng/ngx forks) or a different software with the same name?

Because you can set up mail inbox processing in that software but I’m not sure how that connects to forwarding to multiple different mailboxes, or vice versa profit from merging multiple forwards.

I was running both for quite a while a few years ago.

I found Loop Habit easier for ‘quick entry’ style record keeping, whereas track and graph made it a little more fiddle. On the other hand it allows quite some freedom in what you track and how so it might be better adapted for more in-depth mood tracking.

Nowadays I use Table Habit and have a simple positive mood and negative mood ‘habit’. If I don’t check either the day was neutral. If it went extensively in one of the directions, you can long press and add a custom value.

It’s definitely simplified but that also makes me much more likely to actually keep tracking over longer time frames.

Lastly, another tracker that I’ve used in the past and that is specifically geared to tracking events and emotions is dailynomie. It’s a little harder to set up but should be possible to work as a PWA on android. Loved the simplicity of its data model but the official app shut down in 2023 and I don’t know how maintained the oss version is nowadays.

That’s brilliant news!

I know this is still just very basic federation with mutual following of users, etc. - but having a working mvp implementation of this makes me really hopeful for the stepwise introduction of more and more federation goals to come to fruition.

Absolutely love it! I think this might be my new favourite of yours.

Just throwing another alternative out there which I’ve been very happy with: lissen - ‘Clean audiobookshelf player’.

Has streaming/downloading for offline listening, sleep timer, custom bookmarks and custom speeds. Just the browsing functionality is at best ‘functional’ (nice search but if you don’t know what to listen to next just gives you one long list sorted by title)

I am one of those happy taskwarrior users. For more advanced recurrence syntax in taskwarrior there’s also the relatively new nautical extension. It essentially splits recurrence into things that recur based on calendar/clock events (e.g. every first workday of the month, every second Sunday, etc.) or based on previous completions (e.g. do something 3 days after last time, or every x hours after last doing it).

It’s pretty neat and functional but, fair warning, also early on in development and changing quite radically through its versions still.

While I’m a big proponent of version pinning your critical services, if you’re running stuff in docker swarm shepherd is a solid service updater for the less critical things.

Couple of days ago ‘honey I shrunk the vids’ was posted here, a very lightweight tdarr alternative. https://piefed.ca/c/selfhosted/p/571213/honey-i-shrunk-the-vids-mr-universe-edition-v1-0-5

But I haven’t tried it - I’m not sure if it’s perhaps gui-only, and it seems to be AI coded to a certain degree, so YMMV.

Conversations is working very well on my phone as UP distributor.

On a more personal note, still eagerly reading your updates here every couple of weeks. Sounds like it’s shaping up nicely!

Good luck with the api wiring to let people control the forges directly from the interface for 1.0 - if you ever need a tester for forgejo integration you can hit me up.

While the comment was obviously a bit rude, perhaps you can take something constructive out of it. I see that you have a nice compose file prepared as an example with postgres, and a curl into bash for ‘automatic installation’.

Many feel a bit uncomfortable with doing that - especially if the script ultimately just sets up docker containers again. Maybe you could add a ‘quickstart’ section there which just consists of a single docker run command setting up the sqlite version.

That way you have the best of both worlds and both people wanting a production setup example and those just wanting to quickly try are served.

Very happy you had fun making the little script! One thing that will become important pretty quick if you continue making these scripts is that it’s almost always better to wrap your variables in quotes - so it becomes yt-dlp -x “$a”. It’s okay here but if you ever paste something that has a space in it, this will keep it together ‘as one’.

If you want to expand your knowledge with this, some fruitful paths to go down are the following:

• can you find a way to download multiple urls one after the other if you paste them all at once? (Multiple arguments)
• can you find a way to ask the user for these multiple urls one after the other? (loops)
• and can you find a way to have it ask until you hit enter without a url pasted and only then it starts? (conditionals and test)

The last one is already quite a bit advanced but if you can do that you have enough of the ‘programming’ basics of the shell down to a degree that you can create many little helpers like this with ease.

Of course don’t feel forced to do any of that - if you’re happy with the improvement as-is, that’s all you need to enjoy the fun of Linux!

I have long had a switch to pangolin for my homelab services on the roadmap. Can you explain what you mean with internal connections making more sense?