what features would you like to see in fedibooks?
things i'm definitely doing:
. customisable post frequency
. multiple bots
. bots learn from multiple accounts
. edit name, profile picture, etc. from fedibooks
things i'm probably doing:
. misskey support
things i'm not doing:
. patreon exclusive features
. different types of bots
the source code (what little there is so far, anyway ;3) is at https://github.com/Lynnesbian/FediBooks
here's the bot edit page so far, it looks pretty gross i know
are there any options that seem missing? there are some other options not pictured here, such as turning off replies, banning given words, or changing the profile picture
i don't want to overwhelm the user with options but i don't want it to be too limiting either
before i shut down my free bot hosting service, i was hosting 108 bots. that means my server had 108 open connections listening for replies at all times. i don't know how many it can handle, but the answer's at least 108
i have no idea how many people are going to use this, though, so it might not go so well. each reply listener needs to be its own process, too... i might have to either get a separate server just for this or upgrade my current server to the prohibitively expensive next tier
in other words, when this goes live, i'm kinda hoping for people clicking on those donation links so i can upgrade the server!
this is by the people (henlo) for the people (you!) so if there's anything at all you'd like to see, let me know!
what do you want in an ebooks bot? what annoys or disappoints you about the current ones? if you could program, what custom functionality what you give your own one? what do you like or hate about webapps (like mastodon and gmail) that could apply here? what is your favourite breed of rabbit?
the more hardcore lynnesbian fans may note that fedibooks looks rather similar to curious greg (no not , i mean https://cg.lynnesbian.space)
i did take inspiration from that design, which in turn took inspiration from mastodon's web UI. however, i made a few changes, most noticeably to the buttons, to give it its own unique identity. i also decided to go a little bit brutalist (and i really do mean a little bit) by leaving everything with sharp, square edges, flat colours, monotone icons, and so on.
who says programmers can't design frontends?
you may also notice that fedibooks used to be a different project. i've abandoned the old fedibooks and moved to this one instead. in retrospect, the old fedibooks concept was way too ambitious. it was around the time i was considering how to create a simple UI for displaying complex decision trees that i realised i was in too deep.
why create fedibooks? (long)
I've been working on ebooks bots for a long time. My first project, mstdn-ebooks, dates back to October 2018. It's a much less ambitious project than FediBooks, but maintaining and developing it has still been a formidable undertaking. It's changed a lot since the initial version, and has worked very well for a long time.
Using mstdn-ebooks is nowhere near as easy as using FediBooks. There's a long guide you'll need to follow, and you'll have to install some programs like Python and possibly Git to get working. It runs on your own computer, meaning that if you put it into sleep mode, it stops working. Updating it to the latest version is also your responsibility, and the error messages it gives aren't exactly user friendly.
To help with this issue, I decided to create a Patreon where I offered a $2.50 per month tier to host up to three bots for you. I would take care of configuration, maintenance, updating, et cetera. This worked pretty well up until recently.
A new change in Mastodon and Pleroma adds something called authenticated fetches. Reffered to as "authorised mode" by Mastodon, this new feature prevents unauthorised instances and programs from accessing the API. In other words, if instance A blocks instance B, then instance B won't be able to see instance A's public posts anymore. However, this also means mstdn-ebooks can't see your posts. This can be fixed, however, by requesting permission from instance B. So what's the problem?
I would need to ask every single person I run a bot for to click a link I send to them, and then send me back the authentication code they received. This is messy and annoying, and greatly tedious. Furthermore, mstdn-ebooks' code base is starting to show its age (I used to be even worse at programming, believe it or not), and I thought this gave me a good motivation to create a new project.
When FediBooks needs permission from instance B, all you need to do is click "authorise". No copy and pasting codes and sending them in a Patreon DM. This is much cleaner and easier for the end user, which is what this is all about!
FediBooks is one of the biggest projects I've ever taken on, and it's been wonderful working on it. I hope you enjoy it!
when you ask me a question like "why pour all this time and energy into creating a free service" the answer always boils down to "i want to make people happy"
that's all there is to it really. my current dream job (not that i'm really capable of working) is a computer science lecturer, because i've seen how dry and boring compsci lectures are and i think i can do better. i want people to learn and have a good time doing it!
if i'm making other people happy, i'm almost always happy! i just need to remember to not let that get in the way of my own health.
known issue: fedibooks will stop working after 4,294,967,295 people connect their accounts
looking over this SQL is writing is making me feel very Powerful
i could change fedibooks to only allow one user per instance and all it would take is a single word
i could cause privacy issues and wasted disk space by just deleting one rule
i could change a single column's data type and break pleroma compatibility
a difficult thing to determine is how high should i set the post length limit
my instance has it set to 65535 to ensure i never run into issues, masto defaults to 500, pleroma defaults to 5000
i guess this is supposed to be a microblogging platform so it doesn't make too much sense to allow them to be overly large, and that does technically expose me to some jerk filling my DB with posts that are gigabytes long
65535 it is
the SQL is (probably) finished and it is beautiful
@lynnesbian If you need to save on bytes, you can call it an @
or a full account name since that's what webfinger calls it I guess
@lynnesbian does cg.lynnesbian.space take input anywhere?
(wasn't going to pwn it, just wanted to see what it was)
@lynnesbian OH you took it down before i checked
i thought it would have 404ed, not redirected
though in hindsight i guess it'd be silly to post about how you will, in the near future, close a security hole
@lynnesbian call it your fully qualified user name (or FQUN) so that everyone who visits the page is equally confused
@lynnesbian very good idea
sometimes you do need JS, but whenever it can be avoided, it should be - just to make sites run better
question thou, how are you doin API calls?
@calm you mean interacting with the masto API?
the whole site runs on flask (python) so i'll be using that
@lynnesbian huh - so client side scripting is done in Python?
or is Flask in the backend?
- wait in that case is Python integrated with the frontend AND backend?
@calm flask is the backend ;3
on the frontend, there's not really much that requires interactive JS. there will be a few things, like the chat window, that require JS
@lynnesbian oooo, so if I do something on the site that requires an API call, then Flask handles it directly?
@calm well all the stuff happens on a schedule, like you set up a bot to post every five minutes, and the server does that in the background. it handles replies in the BG too since user interaction isn't needed for that
the chat UI requires live interaction because you're "talking" to the bot, so the server and client communicate with JS in that case
@lynnesbian ohhhh, so for live interaction you use JS?
so if I, for example, press a button that makes a call to the Masto API, what would happen exactly?
@lynnesbian lynne your sentence seems to have fallen on the floor and gotten all jumbled up
here, I fixed it for you:
JS is bad and I'm nice
@lynnesbian Let's just implement the relation types as a property of that foreign key in the database engine and call it a semantic database. Relational databases will be hip again!
@firstname.lastname@example.org's anti-chud pro-skub instance for funtimes