If you don't put a whopping 1MB of JavaScript in your website, you're doing web development wrong.
/s
If you don't put a whopping 1MB of JavaScript in your website, you're doing web development wrong.
/s
See the other reply: seems that it was caused by Lemmy backend only fetches the first 512kb of the HTML, meanwhile YouTube puts bunch of JavaScript in the beginning of the HTML
Here's the fix that was only merged couple days ago apparently, https://github.com/LemmyNet/lemmy/pull/5266
Crazy, right
Also, this is a good tip in general:
You can't fetch title and thumbnail for YouTube links that starts with youtu.be
. This applies not only in Lemmy, but other websites too, for example Twitter (even before Elon's era).
This is, IMO due to the link doesn't really contain HTML and OpenGraph metadata. It only redirects you. If you try cURL-ing the youtu.be link, you will see that it doesn't have a response body, only some header including Location
header that will tell you where to redirect. The response code itself is 301.
I'm noticing that YouTube video's thumbnail and title can't be fetched somehow. In theory this should work, right? See the post and my comment below:
So I take a peek at the Lemmy's source code.
When you see a thumbnail and title of a URL (including YouTube video links) on Lemmy, what happens is that Lemmy server fetched the website HTML file and get the og:title
and og:image
(this convention is called OpenGraph protocol ^1) located inside <head>
HTML tag.
For example, for this video https://www.youtube.com/watch?v=MbYuEEU5e50, the OpenGraph metadata inside the HTML looks like this:
<meta property="og:title" content="Flume - Lose It feat. Vic Mensa">
<meta property="og:image" content="https://i.ytimg.com/vi/MbYuEEU5e50/maxresdefault.jpg">
Try opening the "Create Post" page, then enter URL of a YouTube video. You might notice that a loading spinner will appear. During this, the frontend of Lemmy is actually creating a request to /api/v3/post/site_metadata?url=[insert your URL]
. When the backend side of Lemmy receives the request from front end, it will run this code ^2 to fetch the HTML and parse it to find the OpenGraph metadata. If the metadata is found, the backend will return it to front end, and you willl see recommended title and the "Thumbnail URL" field should be filled.
In theory, this should work fine, regardless of any website. I personally tried to view the HTML code of a YouTube video and the HTML file indeed contains OpenGraph metadata.
But it doesn't for YouTube videos in Lemmy somehow. And I'm not sure.
I'm suspecting YouTube thought that Lemmy servers are bots and because of that instead of receiving the HTML file of a YouTube video, instead it received a HTML file of a captcha page. I think I'm gonna try contacting admins of my instance to see whether YouTube URL works or not.
It's because Lemmy can't fetch YouTube thumbnails. I swear it used to fetch thumbnail and title automatically before, hmm
Oh wow I completely forgot lol
i kinda want to upload all my shitposts gallery on PeerTube but i can't even find a single instance that is open for registration lol
Let me give you my opinion, specifically as a React developer, if you don't mind.
And let's be clear: I intend this to be a constructive criticism. I hope you understand and don't take it the wrong way.
To be honest, I don't know how good or bad federating one-way is. This is more of a "people" problem rather than a technical problem.
But, to be honest, what I am bothered by, is the fact that the website doesn't give an attribution in the UI about which instance certain users are from and which instance certain certain community are from.
Take a look at this post: https://clubsall.com/posts/theyre-trying-to-charge-luigi-with-terrorism-imagine-that-qfF82
The UI says that the post was posted by u[slash]BytesOnBikes
. If I didn't know better, I'd have assumed this was from a user from clubsall. But if you click the username, you realize that the link says u[slash]BytesOnBikes[at]slrpnk.net
. I think this would be confusing as a user. What if there is the same user under the username BytesOnBikes
from clubsall? At least if you include the instance name, user would know right away that both users are different. But if you didn't include the instance name, I feel like this can be abused to impersonate user. This is a bad thing to happen to your website, don't you agree?
Now that we both understand that lack of attribution is a bad thing to clubsall... What's stopping you from adding an instance name to the username? I'm sure the app has a way to know which instance certain users are from. From what I gather, I feel like this is as easy as appending a string in the code.
I haven't even talked about the community name on the UI. Or the ethicality of misleading attribution.
I don't think people liked the idea lol.
I actually liked kbin/mbin. I used it before moving to Lemmy. I just can't code in PHP (and I have had some trauma using it while doing internship)
My Lemmy instance actually can't fetch YouTube thumbnail, so I actually have to put it manually lol.
See http://www.get-youtube-thumbnail.com/