When we showed Akkoma to some initial users we were launching with back in November, they really didn't like it the native frontend. But the one thing they liked, is the one thing that Calckey IMO does even better, the emoji reactions.
I had a previously undiagnosed heart condition that combined with the right circumstances was going to lead to the heart attack I had. I’ve been heavily encouraged to avoid getting excited, stressed as it can cause another as my heart is currently in a fragile form. Unfortunately, talking about my own health kind of stresses me too now, so I won’t discussing it very much.
Thank you everyone who sent me kind words, I very much appreciate you. ❤️
#UFoI For the moment, all instances are hidden from the website; in the best terms, people were quick to find issues with instances that have not gone through the UFoI vetting process (after all, we still haven’t finalized the constitution) and using that as an excuse to find faults, contradictions etc. Sign ups are now available through a Google form, or direct contact to maintain privacy until we officially launch.
On the upside, perhaps this comes down to a bit of confusion of the professionalism, the site does look finalized and some think this is launched fully, which just goes to show the amazing effort you’ve all been doing. ?
Awoo.fyi is apparently blocked on the fedifence.social block list and I am genuinely upset about this whole process and being slandered with no recourse. According to their rules:
> Who is on the blocklist? > > Mastodon service providers that knowingly host and deliver content that is categorised as: > > Racism or advocation of racism > Sexism or advocation of sexism > Discrimination against family status, gender, gender identity or expression, marital status, sex, sexual orientation, native language, age, ability, race and/or ethnicity, caste, national origin, socioeconomic status, religion, geographic location, or any other dimension of diversity > Xenophobia and/or violent nationalism > Hateful terms, symbols, imagery, and shorthand > False or misleading information that may cause or lead to harm > Malware, DDoS, or exploit code intended to disrupt the network
The problem is that I am not knowingly doing any of this. The server itself is configured aggressively to flag offending materials for review, remove nasties from timeline etc. It even rejects posts that contain offensive racial slurs. I have asked my moderators and they haven’t had any notable content show up either for the last couple of days. The content being filtered and moderated out, which is much of the above is further illegal where the server is hosted.
We have an aggressive Terms of Service that pretty much disallows the content described above at https://awoo.fyi/about/tos
I went through their process to get delisted, which was to visit their Github link and post an issue. If you check the attached screenshot, we did. If you visit our issue, it has been removed or hidden with the entire repository at https://github.com/FediFence/fedifence/issues/12
I have included screenshot of the repository commits before I noticed it disappearing, see last screenshot for image.
I think it’s more, what’s your e-mail address? Then there is nothing else really to trade between people.
Ask people what’s their handle on Fediverse… To which people get confused and then you ask if they’re on Mastodon. Then, maybe. Someone then gives their user@instance, some give just the user without instance and then I’ve had some users give me their e-mail…
I believe that confusion is mostly caused by just the login process (let’s just start calling it the fedi handle or something). The fact that to login to your account, you use an e-mail address, or just your user without domain on most instances. Compare to say Twitter where you can use your username, or most free consumer e-mail providers ask you to enter your full e-mail address etc.
“whereas I think the iOS way was only with registering custom URI schemes.”
Yeah, it usually requires having JS trigger it on the page but I didn’t think it worth mentioning.
“It may be a big risk to trust one entity to host said address, unless they’re someone universally trusted and heavyweight enough to keep it up for near-eternity.”
My overall thought is that ActivityPub was standardised through the W3C and they’ve traditionally hosted things like XSL Transformation schemas among other things, they could be an option; IANA also set aside specific domains for certain technical demands and maintain them which could be another entity that could be viable as an option.
“The “Web-based protocol handler” would be a ‘web native’ way of handling it, without requiring any installed app. In fact, it can be both: if there’s registered handler for ‘web+activitypub’ (for example), it could present that URL, if not, then present the landing page handler (which Android clients could intercept and handle).”
Kind of, Android URL app registration requires a particular domain and can’t just be applied to every domain with a certain URL structure. Using apps on Android feels very broken for fediverse links currently (some actually register A LOT of instance URLs to get around this).
But again; I don’t think that this one single facet is the significant reason behind people not understanding how to use the fediverse, I do think it creates a very unpolished experience though and that’s definitely something you’ve clearly identified.
“…A delete & redraft later: it would be nice if Soapbox had a Markdown preview… and not forget that a post was set to Markdown syntax, when delete and redrafting (which gets unset).”
Markdown wise:
Totally agree on the markdown preview
Markdown when converted to plain text really should could up wioth a nice ASCII alternatives, so we don’t need to manually add things like quotationmarks for Mastodon users.
I changed my Settings to always compose in markdown because I keep forgetting.
“One difference is that Twitter doesn’t have to educate it’s users about federation since it’s centralized, so there will always be that requirement “
I strongly disagree because Fidonet and E-mail don’t need to teach its users they’re not centralized. The UX and UI don’t need to explain it for users to use it effectively.
I see most current issues with the Fediverse to be a UI or associated UX issue.
“As for interactions starting from a remote server, some of that can add unnecessary complexities and slower experience (e.g. the ‘remote follow’ workflow of typing in your WebFinger ID, authenticating, action confirmation, etc). There’s not a lot available to simplify it without expecting browser extensions or similar. “
I certainly agree this particular problem isn’t a trivial issue to solve, mainly because it requires adoption of a solution across different implementations.
“The only thing that could be done is designating a protocol handler URI scheme, such as ‘web+activitypub:’ for remote actions”
Technically speaking, don’t even need to do that. Just have generic specific agreed upon webpage domain for this. Phones apps can intercept that URL, non-app aware browsers landing on that page on the other hand get a normal webpage that can simply accept your username@instance (and maybe offer to store it in local storage so you don’t have to reenter it next time) and do the necessary forwarding to the appropriate instance for an action.
I don’t think though this is where users are immediately running into problems. I think there are a lot more basic problems like content discoverability, which isn’t a problem on stuff like Fidonet.
I frequently think about a situation a couple of new users on this instance complained about recently with me. You want to talk about a particular topic? Well, there’s a group of servers over here that cover it, they’re the standard, except, none of them are accepting registrations, they’re not on public relays, the admins on those instances don’t want to join a relay or don’t respond. They’re federating with your server, but equally, you don’t know anyone on that server and vice versa so discovery is difficult. They locked down the ability to see their feeds, so you don’t even see who you could follow remotely. This is really bad and I think the couple of users that complained about this have given up on the fediverse entirely since they haven’t logged in, in a week now.
I’m not too surprised about the confusion, I’ve seen people linking documents like these in response, or having large drawn out guides which for me just feels very unfriendly.
You don’t need a guide to use Twitter. Approaching this as a guide issue is the problem. Very clearly it’s a bad UI/UX problem that most people are experiencing.
I’m genuinely curious how we could improve it in Awoo.
“if you use pleroma you can disable caching entirely or just for domains that you don’t want cached for obvious reasons like pawoo/baraag.”
You mean pull media directly from their domains?
What I’m discussing with Pleroma derivative doesn’t do any caching of media on the server. The configuration I have just proxies the content when requested and Cloudflare does the caching.
“Meanwhile I’m running this single user instance on an old sparc from 2006”
That’s cool TBH. Considering those things can’t run Minecraft servers without hosing themselves quickly, I’m curious what the load of running that instance on such esoteric and old hardware?
Our instance gets around this problem by not storing content from other servers on S3 (our content is still stored on S3). Instead the requests are proxied through our instance, but we have Cloudflare sitting in front of the proxy. On Cloudflare, we have rules to enforce strict caching policies implemented for any valid content served.
“But I’ve gotten quite a few accounts that signup… And then do nothing, no upload of an avatar, no post, no bio… Nothing. “
My own experience with the few people I have been able to ask on instances I’ve ran is that they logged in, took a look, didn’t find anything they wanted, figured they’d look again later another time and then didn’t care anymore later or forgot.
I’ve also had a few admins from other servers create empty accounts as you describe, but they follow a bunch of accounts back on their server to kind of force them to show up in the servers’ feed.
In a good setup, the compute VPSes getting trashed should not matter, because you should be able to deploy a new VPS and all your data is stored else where. Like database server and S3 storage. Of course, it then becomes a matter of ???
We decided to go with a different provider but still using Cloudflare for caching in front of the S3 object storage, costs more though.
Originally attempted to modify the code for this. However, some ACL things were baked right into the dependencies and from my rudimentary knowledge, it looked like the way forward was to fork the dependencies.
I wasn't excited about the prospect, so I then looked at setting up a reverse HTTPS proxy that would strip the HTTP headers we didn't want. It almost worked, except the API wanted responses back; so, I stubbed some resposnes and then it got too complicated when not all stubbed responses worked quite right.
It's not really a bug, but it would have been nice if Cloudflare just didn't throw any errors and just did enough to pretend/mock it supported those headers instead of throwing blatant error messages.