@abucci Thanks.

Though, being fair with myself, I think they wanted to expand to more regions, but I guess that if I tried to setup a company to offer something like jmp.chat, I'd have to track my users, (at least know their identificative documents), as that's written into law here in the EU.

@abucci Are there any (relatively easy (not having to implement it yourself)) alternatives to jmp.chat?

If there aren't, could you please mention other alternative that there are? (even if you have to do some manual setup with some providers) (I guess this would be a possible oportunity to start a company that offered jmp.chat's functionality for other countries, (I guess having different companies that are managed by different people for the service of different countries (the same as jmp.chat more or less), would encourgeage competition, but not sure))

Asking because I live in a different country, so even if I did register to jmp.chat, the system would probably not let me send messages and or calls to it, as it'd be a crazy expensive call or message, right? And even if I could, calling that number would be expensive for family and/or friends, and/or other people.

@prologic Maybe making some kind of video viewer and downloading the youtube channels you'd like your kids to be able to see.

@prologic Nah, it would over-complicate things, but I was thinking about maybe creating some way for users to easily create the website, like creating a specifically named container or trying to bind port 80, (maybe allow subdomains too, I don't know really)

@prologic Thanks, but I (personally) think setting up services directly is easier.

@prologic Yep, I mean docker exec into a existing container.

@prologic I can't exec to a running contaier, which seems odd.

@prologic How can I host a webapp using the dogfood system?

@abucci You need to install the docker command, as the script you are running calls docker a few times.

@prologic I don't know, I can check in a few hours when I am able to connect my virtual machine to the internet (I am updating the os, and will reboot tomorrow (today in a few hours (UTC+2))

@prologic I mean appart from having ssh cas.run tui being able to be able to ssh cas.run for each action having the TUI be a easy way to access those actions.

@prologic I like the SSH interface, maybe adding a few commands (like how do I list the created docker creds?, you could later create a website management too (all of the website management would call the same code as SSH, and if you make a SSH ui, make all it's functions callable directly and easily from normal ssh (to automate things)

@prologic And maybe adding a flag to be able to use the podman command instead of docker, (it'd allow also changing it to some other provider?)

It's nice you got it hosted rn :), I might try it later on another host where I probably have docker, (I don't have docker on my local computer, so I will probably install docker on a VM, and copy the shell script, and try to use the docker command to run machines on the other host, (would that work?))

Also, do you mind if I try to break outside of the docker daemon?, (If I were to get out of the docker daemon, I would inform you privately on how so you can fix it properly)

@prologic I guess it'd be nice to not require to have a github account when you make this stop being alpha-level software right?

@prologic I don't know, I mean the service you mentioned from the message of the fork thread reply I made...

@prologic Could you explain why Docker's CaaS system failed?

@prologic Could you link to some blog post or some explanatory of why the y failed?, this could make this new service also fail. If it were to be setted up like a container there'd be a few restrictions and you'd be billed for all the storage of the container, and the resources used by it, but that's like all the cloud computing pay-as-you-go plans...

@prologic It solves the issue of finding users from inside twtxt, but you'll have to find registries which is in itself a chicken-egg problem. It also makes it easier to implment web twtxt clients, as they won't have to make that many requests, just one to the registry, instead of having each web-client make a request, but then you lose part of the features, as the registry would have to implment things outside the spec to be able to allow follows, and you seeing them...

@prologic I am saying that you could create your own registry for yourself, your own use, so only you would use, but yeah. a) Yeah, there could still be registries related to communities, so making a follow new users from registry feature might be nice, but I don't know really. b) You can register to as many registers as you'd like c) That's the same problem as right now technically, we could create registries that are related to specific fields, like tech (many tech registries), housing and all that, but I don't really know.

@movq Yeah, I guess then I'd have to make some system to tailor my requirements. So yeah it'll open up the gates for spam I guess, depending on how it's configured.

@movq I think subscribing to the tweets endpoint is a better idea, as accessing the users endpoint can be done easily if wanted, and would be pretty easy to make a short script to subscribe to all users from the registry. So parsing the tweets endpoint (could be the mentions endpoint if wanted, any endpoint that uses the format as the tweets endpoint would be valid) and adding another parameter to that specific follow, (maybe a special 'registry' user, (I think it should check if the user starts with registry, to be able to have multiple registries), and then jenny would just check the registry and handle the messages using the data in those endpoints, changing the user and twtxt url acordingly.

@prologic I guess you could create a registry of your feeds, so that when someone starts following you, it'll be added to the registry. So you can get pings or whatever you want, it doesn't have to be a centralized registry if it's yours.

@movq My idea was to support this in jenny: https://twtxt.readthedocs.io/en/latest/user/registry.html#registry

@movq My idea was to somehow get the twtxt.tilde.insitute to be able to be added as a url, so you'd be able to get posts from multiple users instead of only a single one, but I think that might come with some issues, so yeah better manually populate the urls from the users endpoint.

@movq Oh, I thought that wasn't a feature, if it is, sorry.

I am also thinking maybe adding a getwtxt-ng feature (where the follow list is stored on a separate service), might be nice :P, I might try to send a formatted patch trough here, but I think it's best to use email.

@prologic I will try to keep in mind.

@prologic Is it fine if I make the logs that only contain path and useragent to be able to be seen to all members of my pubnix (anyone can become a member), or should I make it to be able to only be able to see your twtxt logs (by a bash script that runs as root via sudo, that greps for your username in the file).

Yes, I know Baïkal - baikal exists, but I'd have to or a) Register users manually (outside of sso), or b) Create registration pannel that would connect to the api and register the user.

I also forgot to tell that I'd like to be able to somehow 'share' events, even if I have to use some kind of webui or api.

@prologic I guess that'd be possibly true, except he was looking at logs I guess from the webserver.

@prologic I asume they report it to one of the ASN's abuse addresses.

@movq I was able to replicate it in sway. No problem ⌨️

@movq That's pretty weird, I have been able to replicate this using two keyboards. You could use interception to make those two keyboards act as one.

See section 2.2.5 from this page: https://wiki.archlinux.org/title/Interception-tools, for more information.

@movq Shouldn't it just work as it should be recognized as two HID devices, just that they only type half the keys?

@prologic I was testing jenny and didn't make sure I commented the code that actually sends posts, (I removed those posts from my twtxt file alredy), so I mentioned a few people with single-letter messages

I also thinked about sr.ht, but I think sr.ht has some issues, as you still can not, well make the repos from the cli, I think I'll need to make my own managing script for mlmmj or something, maybe even make my own mailing list manager, (probably forking from mlmmj as that's working fine for me)

@prologic Does it work normally now? I think I modified it, but I think yarnd caches messages, so you might need to update the cache?

@prologic I see the threadings properly on my email client, seems weird (I used the fork-conversation feature so maybe that's why?)

I have fixed the weird tags thing, I removed it from the twtxt manually, I don't think it'll affect anything? I wrote a patch for jenny (twtxt to mbox converter), and I did test it in a sense in that messsage, because jenny wasn't making RFC compliant emails and aerc in newer versions needs the emails to be compliant, else it doesn't seem to be able to do threading.

@prologic What do you think about using gitolite and cgit, with a mailing list for patches?

I don't know if I should set it up, I know that for my personal projects it will be sufficient, but I don't know how well it'd scale with a pubnix-like system. I know the linux kernel uses gitolite but I don't know, as the use by us would be completly different, it'd be more suited to giteaforgejo, (because they offer the ability to make organizations), but I also dislike not having the gitolite ui in gitea, so I might think about making custom scripts to let you make projects (that are outside your user namescope, but assigning you privileges, (I think I'd need to or a) make the script directly add the projects, thus making me git pull the gitolite-admin repo, but I'd alredy have to do that as I'd use sskm, or b) Having the requests come in to my mailbox and I manually add their respective config block to gitolite's config in the gitolite-admin )

@prologic 🤣🤣🤣

@prologic Seems like github works for me.

@prologic I think twtxt.net/yarnd don't generate a proper markdown output, it looks like it put ![<link message>](<link> "<link message>") but maybe it's because that's another syntax for something different. I can't tell.

@prologic Shouldn't it be as easy as remembering each user and checking if they are constantly asking for the resource (maybe if they check at least twice in a week)

But I think that then the issue gets into having correct thresholds to detect follows properly from just checking what the user's doing.

@prologic Maybe it's because it's not surfaced to users completly. I don't know, and I get completly out of touch as I haven't yet figured out a way to sort messages as a mailing list thread in aerc with the weird tags that jenny adds.

@prologic Then it's weird that twtxt.net didn't notice my requests to your feed, as I am following your feed in jenny and my anounce setting is set to true

@prologic Yarn seems like it does not support the protocol for discovery that you mentioned.

• mentioning (not posting)

@prologic It seemed to not care that I where posting to you, as my user-agent I think contains my username that was weirding me out a bit.

@prologic Because yarn doesn't seem to work properly with client discovery

That's the reason I want to let me manually add my feed to yarn services. Also, do you have any recommendations on how to implement this using nginx logs?