freamon

joined 2 years ago
MODERATOR OF
[–] freamon@lemmy.world 2 points 2 weeks ago

Unless the person replying on Mastodon removes it of course (to save themselves some characters and because they don't recognise it).

There doesn't seem any functional difference between your comment (which implicitly Mentions the community), and the original post on thecanadian.social (which explicitly Mentions it). Screenies from activitypub.academy show that replies to both include both Mentions pre-filled.


https://thecanadian.social/@mike/116234121499688717 didn't remove the Mention, and so the comment made it to Lemmy. https://mstdn.ca/@cass_m/116234270845317774 did remove the Mention, and so it didn't.


Specifically for Lemmy btw, the reply at https://mstdn.ca/@cass_m/116234270845317774 demonstrates another problem - that instance requires a signed GET request for its actors, which hasn't been fully implemented AFAIK.

[–] freamon@lemmy.world 1 points 2 weeks ago

Already submitted to this community, a few minutes before yours, by the original poster: https://lemmy.ca/post/61609680

[–] freamon@lemmy.world 1 points 3 weeks ago (1 children)

Just come across this myself.

Long-running fediverse instances will likely have old copies of established aussie.zone actors, but federation will be broke for newer instances and/or newer actors.

Verifiable by doing something like:

WORKS:   
curl --header 'accept: application/activity+json' https://reddthat.com/u/lodion

DOESN'T WORK:   
curl --header 'accept: application/activity+json' https://aussie.zone/u/lodion

I chose reddthat.com because it seems like it has similar protection, but allows for AP requests. (other instances use non-CF solutions, but I imagine they've had to make similar tweaks).

It's been a couple of days, so it might be worth some pings:

@admin@aussie.zone

@lodion@aussie.zone

[–] freamon@lemmy.world 4 points 1 month ago

You can get a rough idea of how big each video is via an activitypub query for each video. For example:

curl --header 'accept: application/activity+json' https://peertube.wtf/w/mhghLtY5dkLguNq5oFB2Ut | jq .

Buried in there is a url entry, and buried in some of those is a tag entry and buried in some of those are details for the size of the video and audio for each upload.

Peertube's video channels have an outbox (similar to how Lemmy's communities do, but not limited to 50 entries), so you can step through that to find the relevant info for everything in your channel.

Doing that for your channel, I got:

Shitpost #2
video: 482 x 480: 2.27 MB
audio: 0.3 MB

Shitpost #1
video: 480 x 480: 8.69 MB
audio: 0.94 MB

Will this replace the internet?
video: 1080 x 1920: 19.26 MB
video: 720 x 1280: 11.9 MB
video: 360 x 640: 5.23 MB
audio: 16.75 MB

Cat
video: 1920 x 1080: 7.77 MB
video: 1280 x 720: 2.14 MB
video: 640 x 360: 0.41 MB
audio: 0.23 MB

If you assume that the size of the audio is ignored, and that lower resolutions are transcoded as requested, and add the sizes for the highest resolutions together, you get 2.27 + 8.69 + 19.26 + 7.77 = 37.99 (which is the 38 MB visible in your screenshot).

The information is available, but it's a pain in the arse to get, so it's probably annoying for PeerTube themselves to show (a brief look suggests that the API response doesn't provide it, so there's nothing for the web frontend to display). It's also possible that they may drop the higher resolutions for videos with low engagement, so the size of each upload isn't static, which adds an extra complication.

[–] freamon@lemmy.world 4 points 1 month ago (1 children)

Not really.

Gup.pe groups were genuine ActivityPub Groups, like Lemmy communities, whereas these 'FediGroup' things are just Mastodon bots. They're a 'Service', aka the automated version of a 'Person', so they're no use to anyone on platforms (like Lemmy) that can only follow Groups.

The most similar recent thing to gup.pe is https://ovo.st/

Compare:

curl --header 'accept: application/activity+json' https://lemmy.world/c/fediverse | jq -r .type   
=> Group   
curl --header 'accept: application/activity+json' https://ovo.st/club/askfedi | jq -r .type  
=> Group

with

curl --header 'accept: application/activity+json' https://fedigroups.social/@audiofiction | jq -r .type  
=> Service
[–] freamon@lemmy.world 2 points 1 month ago (1 children)

Dunno if you're fussed about this sort of stuff, but this meme was already posted very recently: https://crazypeople.online/post/14691687 (sidenote: lemmy's default UI doesn't do inline displays of avif files, it seems)

[–] freamon@lemmy.world 1 points 1 month ago

Not the first time that bug's reared its head either.

Example from here:

'nsfw': post.nsfw if post.nsfw is not None else False

A meme about PieFed half-arsing database migrations might not be funny, but would at least be valid, and less wearisome than OP's post.

[–] freamon@lemmy.world 1 points 1 month ago (2 children)

I've contributed code to PieFed in the past, but nothing recently. If someone comes across something I've written and finds it amateurish, then that's a reasonable assessment. There's no need for you to delete your comment, as I'm not a fan of features over fixes approach either.

The "People's Front of Judea" remark relates to a Monty Python sketch from the The Life Of Brian (youtube link) - it's a swipe at leftist infighting (swap out "The only people we hate more than the Romans is the Judean's People's Front" with "The only site we hate more than Reddit is PieFed" I suppose).

[–] freamon@lemmy.world 17 points 1 month ago (6 children)

This has come up before. Hopefully you're just not understanding the code, rather than deliberately misrepresenting it to others. Even a casual scan should clue people in to the fact that the linked function isn't concerned with federation blocks (the same list that 'enoughmuskspam' is in also contains 'memes' and 'piracy', which every PieFed instance has without any overrides required).

I'll copy-paste my comment from last time (I can't link to it 'cos is was in reply to a deleted post). The first 2 paras are the most relevant bits:

The code that OP has linked to is part of a convenience function for admins to add content to their new instances. It can query individual remote instances (e.g. lemmy.world), or it can query lemmyverse.net, and fetch communities that look to be popular and active.

It’s completely unrelated to routine federation, and doesn’t prevent anyone subscribing to communities that may have those words in their names.

The admin function could potentially be used to fetch hundreds of communities. It runs as a background process, so you don’t know what they were until after they’d been followed. The “bad words” list acts as a safeguard against bringing in things you might not want or expect. One reason is that you may want to curate the first impression you give new visitors, as there as some that will be put off by the “fuck this” and “shitpost that” reddit-isms. Another is that you don’t typically want communities that are disproportionately popular than others (e.g. if you bring in the default 25 communities, and one of is 196, then it completely dominate your front page).

If there’s a particular community that you are interested in (e.g. because you moderate it), using this function isn’t an efficient way to add it. In addition to the “bad words” filters, it will also exclude communities that are NSFW, or below thresholds for popularity and activity. Rather than fetching a bunch of communities at the same time, and hoping that the one you want is included, it’s better to just add it manually (via a ! link or by using the “Add remote community” link) in much the same way as you would on any other platform.

[–] freamon@lemmy.world 29 points 2 months ago

It's mostly annoying me because it's affecting federation - not just the ability for new instances to backfill content, but for established instances to even be able to fully resolve a new post's details before the author nukes themselves. For example, the previous post (titled "Paranoia") isn't available on lemmy.ml or lemmy.dbzer0 because they would have been a few seconds "too late" to fetch the details for the author.

@lawrence@lemmy.world - if you'd like another mod to help out in the short term, I'm happy to volunteer. I realise people in the comments are being a bit defeatist about the ability to reckon with this problem, but whoever this person is should at least have to do a bit more work than they're currently doing.

[–] freamon@lemmy.world 25 points 2 months ago* (last edited 2 months ago) (4 children)

The code that OP has linked to is part of a convenience function for admins to add content to their new instances. It can query individual remote instances (e.g. lemmy.world), or it can query lemmyverse.net, and fetch communities that look to be popular and active.

It's completely unrelated to routine federation, and doesn't prevent anyone subscribing to communities that may have those words in their names.

The admin function could potentially be used to fetch hundreds of communities. It runs as a background process, so you don't know what they were until after they'd been followed. The "bad words" list acts as a safeguard against bringing in things you might not want or expect. One reason is that you may want to curate the first impression you give new visitors, as there as some that will be put off by the "fuck this" and "shitpost that" reddit-isms. Another is that you don't typically want communities that are disproportionately popular than others (e.g. if you bring in the default 25 communities, and one of is 196, then it completely dominate your front page).

If there's a particular community that you are interested in (e.g. because you moderate it), using this function isn't an efficient way to add it. In addition to the "bad words" filters, it will also exclude communities that are NSFW, or below thresholds for popularity and activity. Rather than fetching a bunch of communities at the same time, and hoping that the one you want is included, it's better to just add it manually (via a ! link or by using the "Add remote community" link) in much the same way as you would on any other platform.

 

As you may have noticed, the crawler at https://lemmyverse.net isn't picking up anything from instances on versions 0.19.4 or 0.19.5.

The Issue itself is easily fixed, and there's already a PR for it from lemm.ee's admin, but there's been no response from the lemmyverse developer.

Does anyway have any other ways of contacting him?
If this continues (I realise 2 weeks isn't that long), is anyone interested in forking the code and hosting it on a new domain?

Thanks.

 

(bonus opportunity to brush up on your Portuguese swear words)

 

I realise this is a known issue and that lemmy.world isn't the only instance that does this. Also, I'm aware that there are other things affecting federation. But I'm seeing some things not federate, and can't help thinking that things would be going smoother if all the output from the biggest lemmy instance wasn't 50% spam.

Hopefully this doesn't seem like I'm shit-stirring, or trying to make the Issue I'm interested in more important than other Issues. It's something I mention occasionally, but it might be a bit abstract if you're not the admin of another instance.

The red terminal is a tail -f of the nginx log on my server. The green terminal is outputting some details from the ActivityPub JSON containing the Announce. You should be able to see the correlation between the lines in the nginx log, and lines from the activity, and that everything is duplicated.

This was generated by me commenting on an old post, using content that spawns an answer from a couple of bots, and then me upvoting the response. (so CREATE, CREATE, LIKE, is being announced as CREATE, CREATE, CREATE, CREATE, LIKE, LIKE). If you scale that up to every activity by every user, you'll appreciate that LW is creating a lot of work for anyone else in the Fediverse, just to filter out the duplicates.

 

For anyone unaware, a community's outbox typically contains the last 50 or so Post Announcements - it is retrieved when you are the first person to find a community on a remote instance. It also seems to be fetched whenever an community on a remote instance realizes it's out of sync with the community on its host instance.

Compare:

curl --header 'accept: application/json' https://lemmy.world/c/memes | jq -r .outbox           
=> https://lemmy.world/c/memes/outbox                
curl --header 'accept: application/json' https://lemmy.world/c/memes/outbox | jq .       
{
  "error": "unknown",
  "message": "Record not found"
}

with

curl --header 'accept: application/json' https://lemmy.world/c/linuxmemes | jq -r .outbox      
=> https://lemmy.world/c/linuxmemes/outbox          
curl --header 'accept: application/json' https://lemmy.world/c/linuxmemes/outbox | jq .orderedItems[0]   
{   
    "id": "https://lemmy.world/activities/announce/create/0223f939-aafc-4215-9c20-a3460c967f63" 
    ... (the rest of the most recent post from linux memes)
}

(I came across this randomly, so I don't know if 'memes' is the only community missing an outbox. Others I've tried have been OK though)

 

Hello again,

February's finally over, so 4 Frame memes are old news. Now, for March, memes with rhymes in them are what's required to get Featured in the Community.

This one will be stickied until a new meme with (ideally terrible) rhymes in and [ILPM] somewhere in the title is submitted, and then the new post will be stickied instead (it's a manual affair atm, so it won't be instant)

Thanks!

 
view more: next ›