I don't understand one thing...why do you need to host it on a server? It can be just a standalone app on windows/Linux/mac. I recently just started using yt-dlp and have thought of making a simple app. Should I though? Are there any alternatives in existence of this kind?
> why do you need to host it on a server? It can be just a standalone app
Your Windows/Linux/OSX device, or a VM or container on it, can be the server if you wish. This way the creators don't need to make two UXs (web for the server, native for the other).
A lot of people run thier self-hosted apps on an external server or home lab (or mix/multiple) with those resources being "properly" setup as infrastructure with backups and such, using their PCs/laptops/tablets/phones as relatively dumb clients/caches that can be easily replaced (pulling local content back from the server the old devices synced to) if lost/damaged. This is one reason self-housing that way is popular.
> using yt-dlp and have thought of making a simple app
I've seen a few desktop apps that seem to be wrappers around yt-dlp (and/or other similar utilities). I've not used any of them so can't comment on their stability/reliability/other, but I'd do a search before writing your own from scratch (unless you want to do that for fun anyway!) as you might find what you want already exists.
it also neatly integrates into easy-to-use things like runtipi. I have it on my spare home server, running a tandoor instance, invidious and metube for watching stuff later/offline.
yt-dlp is not even necessary. If just downloading videos from YouTube this can be done with much less code and complexity. I use a tiny C program to do it. Tiny shell scripts work just as well.
What no one is discussing is that yt-dlp can no longer download itag 22 and itag 18 is now throttled.^1 This means yt-dlp has to download an audio file and a video file and use ffmpeg to merge the two.
YouTube ad revenue just missed consensus estimates. Perhaps the accessibility of videos on YouTube is going to get even worse.
1. One will not likely notice this unless one is in the habit of specifically choosing format. For example, some folks have used itag 18 and 22 exclusively for many years.
> yt-dlp is not even necessary. If just downloading videos from YouTube this can be done with much less code and complexity. I use a tiny C program to do it. Tiny shell scripts work just as well.
yt-dlp is not necessary ... if you write a replacement? Ok.
Besides I doubt simple shell scripts will be able to download all videos that yt-dlp can. YouTube does not use the same protections for all videos.
"Besides I doubt simple shell scripts will be able to download all videos that yt-dlp can."
You are free to doubt but the the truth is yes they can.
The only ones that are complicated are commercial videos.
Give me a list of videos you are downloading with yt-dlp and I will show you how to dowload the same videos without yt-dlp. I have been doing so for many years. Unless you are only downloading commercial videos, e.g., VEVO crap. Personally I am not interested in those videos.
I am actually liking this YouTube change getting rid of or throttling video with pre-muxed audio. Now I download only the audio. If I like it, then I download the video and mux. For many videos, there is no meaningful video; it is just slideshows of images or animations. For me, audio-only is faster and more space-efficient. I am avoiding downloading many videos because in many cases the audio is enough.
> yt-dlp can no longer download itag 22 and itag 18 is now throttled. This means yt-dlp has to download an audio file and a video file and use ffmpeg to merge the two.
I found yt-dlp did that more often than not anyway, if left to its own devices on choosing which formats to pick. Though I had noticed it doing that pretty much always recently.
> Perhaps the accessibility of videos on YouTube is going to get even worse.
I think that has started to happen. I've had a few occurrences recently where it tries to get a particular format and fails, so I've had to manually request something else.
They also seem to have started screening non-residential/office/similar connections more aggressively too, two external hosts of mine now get a message to the effect "login to prove you aren't a bot" when yt-dlp tries to pick something up, and I've seen people note that they have trouble with YouTube proper (without logging in) over some VPN services.
This has been the case for a while now. It may be temporary. Time will tell. Not being a video codec nerd, I do not find that itag 136 is any better than itag 22. Files are about the same size. But 136 needs separate audio file which creates need for ffmpeg and, generally, more temporary storage space.
I also realised the same thing yesterday. I tried using a specific tag quality, for a playlist and many of the videos failed to download. So, I just set it to download the best available quality for video and audio.
The question then becomes why not just ssh into that computer or whatever and just use yt-dlp on a terminal on it instead of a javascript wrapper around it.
you host it on the server which has a raid array and is/gets backedup vs a laptop that could get stolen or lost. the point is to have a copy of the videos that can't be deleted/lost so having that on a loseable laptop wouldn't satisfy that requirement
Not the GP but for me personally, no, never. I only sync a couple of text files which I update and reference regularly enough to need them to be synced across machines constantly.
I don't really understand HN's seeming aversion to having a headless server that handles centralized tasks like downloading, storing, and serving files. It's a great model for personal computing with a ton of upsides. As far as I'm concerned all storage on my laptop is totally disposable, and I should be able to wipe and reinstall everything quickly without loss of important data.
I back up the data on my laptop daily to rotating backup disks + there's a local server for git and some other stuff. So yeah, storage on my laptop is disposable. Which I test whenever I get a new one: restore from backup. Works.
Where to? Some sort of server? where you could run some sort of software that would organizeand catalogue your files in a more readily usable format than digging into a laptop backup archive?
