I have said and will keep saying this, Snap continues to be one of the worst things Canonical has tried to pass off as “production ready” at least in recent times, and every time they push more for it makes me consider them even more of an untrustworthy distro vendor.
My personal experience with it is that I had to nuke it off the computers I maintain last time we updated Ubuntu to the new release (unfortunately I don’t control what OS we’re using, otherwise Ubuntu would have been gone faster than you could say “Snap”), because these computers have user homes on a network drive, which isn’t under /home, and snap just flat out refuses to work when that is the case.
It wouldn’t even be such a problem if they hadn’t removed Firefox among other things from their repositories, but as it is right now, there’s no way to run Firefox on Ubuntu without mucking around with PPAs if your user home is not under /home. This bug about that has been open for 7 years!
That said, I think this is fine considering Flatpak is still going to be available in the repositories, just not installed by default. I’m much more a fan of “install what you need”, anyway. (But then again, that’s not the style of distro Ubuntu wants to go for.)
I’m using Pop! OS and basically I only need to use the terminal when tinkering with Wine (because I’m stubborn and don’t use Lutris), launching StableDiffusion or doing some wacky shit like trying to install a compatibility layer for android apps.
Looks like a cool script. I am learning different TUI tools now and willing to migrate some from GUI. This one gives me trouble. I probably did something wrong. When I press m then P - nothing happens. Supposedly, there should be preferences. And the program didn’t ask me on the first run to configure anything.
When I search a video with s - I get in a preview window: zsh:11: command not found: draw_preview
Strange… It still displays the same zsh error and the preferences are not displayed. I even tried to execute the program using bash magic-tape.sh while in bash
I tried on a machine where only bash is installed, and the search window is blank, yt-dlp gives an error that flashes and I can’t read quickly, no preferences appear, again it flashes some text but I cannot read it. Also, the location for the folder is not relative, it has to be in ~/git/magic-tape otherwise it will give you multiple errors.
Yeah, I followed the instructions, created necessary folders. yt-dlp is not the latest version, but the latest from Debian bookworm repositories (2023.03.04). yt-dlp works perfectly to download and convert videos with ffmpeg.
When I try to browse trending yt-dlp says:
<pre style="background-color:#ffffff;">
<span style="color:#323232;">yt-dlp: error: playlist end "0" must be must be greater than or equal to playlist start "1"
</span>
I don’t have zsh errors on Debian where only bash is installed, after searching for a video I got a blank screen with the only option:
You can open the P option by either double-clicking on it with the mouse, or entering capital P with the keyboard. You can get to select the P option through the misc menu.
Do you get any error messages when doing so?
Have you installed all the dependencies of the script (rofi in particular)?
I did not have rofi installed on my headless machine, from the description I understand it is a GUI application. After installing it I still could not get the Preferences menu to appear. Nothing happens when I press P, or more specifically there is a quick flash of something and it comes back to the m submenu
I finally got to my main workstation with GUI, pulled the latest code, installed rofi there and it works now, I sill could not to get previews to show because of the draw_preview: command not found error. But I don’t think rofi can run on a headless machine:
<pre style="background-color:#ffffff;">
<span style="color:#323232;">rofi
</span><span style="color:#323232;">(process:1456958): X11Helper-WARNING **: 10:52:55.600: Failed to open display: (null)
</span><span style="color:#323232;">(process:1456958): Rofi-WARNING **: 10:52:55.600: Connection has error
</span>
I think that apart from not installing the dependencies, both problems that you have faced in your machine (zsh-bash conflict, headless machine) are beyond the scope of this script and need to be examined individually.
At the end of the day, one could easily substitute rofi with fzf itself in the script, they do more or less the same job. I am not certain rofi is your main obstacle here.
I tried to substitute rofi with fzf in vim, and when I try Preferences - it says 😕 Selection canceled…
My main goal was to use this program on my server which has no GUI and I can access it from anywhere. For example, I successfully use newsboat and neomutt remotely. When I saw TUI, I thought I could use that too headless, but you may wanna put a disclaimer that the script requires a GUI application rofi.
Anyways, thanks for the project, I will keep an eye on it if it keeps developing, great idea! While troubleshooting this program I came across ytfzf that also supposedly supports subscriptions and seems like can work headlessly
Don’t get mad, I’ve read. It’s a little unusual for me to see amongst 6 CLI dependencies on GUI one. But even installing a GUI application on a headless machine won’t do any good
The directory structure of the program has been updated. Instead of keeping everything in ~git/magic-tape/, now various files and directories are kept in various places.This way,
the magic-tape.sh is in ~/.local/bin/
the magic-tape cache files are all in ~/.cache/magic-tape/
the configuration text file will be created in ~/.config/magic-tape/
The action selection can be either with rofi, or fzf (if the user wants to go full TUI).This can be configured during the P option of the misc menu.
So if you want to go full tui, you can avoid rofi and go full fzf.
Actual git still worked. I was able to git pullI also figured out a way to launch the script using bash - needed to export $SHELL - the only way it worked. I could not update yt-dlp to the latest version - the latest I could install is 2023.06.22 from the side PPA repo. Official Ubuntu repo provides 2023.03 and pip breaks my system with compatibility issues that I don’t have desire to troubleshoot. I could try in the future downloading their binary but I don’t like when stuff doesn’t auto-update. The preferences worked this time and I was able to save them.
yt-dlp gives me an error every time I try to browse either Trending or do a search. I have a TV with a cross and 1 option to Abort Selection.
I have a question: is the script dependent on having browser cookies? Because I don’t have any of the browsers installed on a headless machine. ANd I think that is what yt-dlp wasn’t happy about…
<pre style="background-color:#ffffff;">
<span style="color:#323232;">Downloading /feed/trending ...
</span><span style="color:#323232;">Traceback (most recent call last) :
</span><span style="color:#323232;">File "/usr/bin/yt-dlp", line 33, in <module>
</span><span style="color:#323232;">sys. , 'console_scripts'
</span><span style="color:#323232;">' yt—dtp'
</span><span style="color:#323232;">Fite py", tine 1008, in main
</span><span style="color:#323232;">File "/usr/lib/python3/dist-packages/yt_dlp/_init_.py", tine 962, in _real_main
</span><span style="color:#323232;">with YoutubeDL(ydl_opts) as ydl:
</span><span style="color:#323232;">AAAAAAAAAAAAAAAAAAA
</span><span style="color:#323232;">File "/usr/tib/python3/dist—packages/yt_dlp/YoutubeDL.py", tine 762, in _ init
</span><span style="color:#323232;">self. _ setup_opener()
</span><span style="color:#323232;">File "/usr/Iib/python3/dist—packages/yt_dIp/YoutubeDL.py", tine 3929, in _ setup_opener
</span><span style="color:#323232;">self. cookiejar = load_cookies(opts_cookiefile, opts_cookiesfrombrowser, self)
</span><span style="color:#323232;">File "/usr/lib/python3/dist-packages/yt_dtp/cookies.py", line 106, in load_cookies
</span><span style="color:#323232;">extract_cookies_from_browser(browser_name, profile, YDLLogger(ydI), keyring=keyring, container=container))
</span><span style="color:#323232;">File "/usr/lib/python3/dist-packages/yt_dlp/cookies.py", line 123, in extract_cookies_from_browser
</span><span style="color:#323232;">return _extract_firefox_cookies(profile, container, logger)
</span><span style="color:#323232;">File py", tine 148, in _extract_firefox_cookies
</span><span style="color:#323232;">raise FileNotFoundError(f' could not find Firefox cookies database in {search_root}')
</span><span style="color:#323232;">FiteNotFoundError: could not find Firefox cookies database in /home/user/. mozitta/firefox
</span><span style="color:#323232;">Completed /feed/trendinq.
</span>
Unfortunately, (or fortunately for me) I have not used YouTube in a traditional way in a very long time. I use FreeTube, Piped, Invidious, NewPipe, VueTube depending on the platform I currently am. And that is why I don’t have any cookies for it. I don’t really care for liking or commenting the videos as just watching them. All the above mentioned programs support own “offline” subscription lists, and most have interest-based profiles which you cannot achieve with a traditional YouTube account.
But yeah, sadly this not for me in its current form, I will join whoever added a comment here earlier about waiting for support of json - based subscription lists - that is, of course, if you decide to go that route, or at least as bare minimum if you could remove the cookie requirement and have “anonymous search and view” functionality. I believe many of us here on Lemmy that are concerned about privacy will greatly appreciate that. YouTube itself, mpv, and yt-dlp don’t really need cookies to watch a video.
I was considering being full “offline”, it is perfectly doable, working with yt xml requests.
The downside is that this process (so far for me) takes longer, gives less info, and mostly, although there is no image support there, is done better by terminal rss feeders like newsboat.
You cannot do everything at the same time. I may revisit this approach in the feature and put up something else.
I tried it like a year ago maybe and it was a very interesting concept, but I couldn’t find what it really brings to the table over Guix or Nix. I’ve used Guix quite a lot since then and found that the separation of packages was afaict very similar, with the added bonus of having a completely reproducible system. Maybe Gobo brings something that Guix/Nix does not, but in my time of testing it I couldn’t find anything.
I have been using Debian for about 20 years now. Server and desktop. But I recently migrated all my server stuff to FreeBSD and I don’t think I will move back. Jails are great and provide me a convenient way to isolate my apps. On the desktop side I will stay with Debian.
NixOS needs what is IMO the killer feature of Arch: the wiki.
Comprehensive documentation on not only the OS but the additional packages that we use is what drew me to Arch, and the thing that makes me swear in frustration whenever I have to use Ubuntu/Debian.
NixOS is an excellent OS that has the promise of being every bit as hackable as Arch, but far more stable. Problem is, configuration is very different and needs extensive documentation to reduce that friction point.
The Arch wiki is pretty distro-agnostic (barring package names and pacman specific stuff). I’ve been distro-hopping for past decade and I’ve always used it as a reference for setting things up.
Have you bothered to research the consensus and what Debian’s new release has? Literally 2 minutes away if you search internet instead of replying. Do not expect spoonfeeding.
Is your password longer than 60 characters? If so, you should know that your actual password was truncated to 60 chars at the time of account creation. You won’t notice on the website because the login webform automatically truncates characters beyond the limit, but most 3rd-party apps don’t do this because it’s an undocumented thing.
Basically, my advice would be to update the password in your password manager to be truncated to 60 characters and then try logging in with the app again.
Oh man, the feeling of solving an obscure issue on the first try without having to ask for any extra hints: chef’s kiss. I’m glad I could be of assistance!
It doesn’t seem like it, RHEL is based on Fedora. Fedora is beta RHEL more or less. But Alma, Rocky and Scientific Linux will all be in trouble as they’re based on RHEL. Or at least that’s what I make of it
You might want to rename it considering that a company named Liftoff Software exists and has a trademark on the the term, “liftoff” pertaining to software.
I’ve tried just about all of the current apps that are available for lemmy over the last two days and so far liftoff feels the most like a usable app to me. To ve fair, I know that a lot of rhe apps I tried are still in alpha or beta, so I’m not trying to talk bad about them, I get it. Liftoff is pretty good though.
The Subscribed feed is showing up empty. Other settings regarding the feed work fine, but the Subscribed feed is broken. Also, replies showing up as double replies, one below the other, once you reply to a post.
And you could make the search menu a bit more… friendly. Currently you have to add instances in order to search for communities in them. Not practical at all, there are hundreds of instances, if I had to add each one in order to search in them, that will just take way too long. Why not have instance and communities auto-discoverable, like in Jerboa (type in search the community you’re lookimg for, results pop up).
About the search menu… unless I’m mis understanding that’s just how Lemmy works. You have to be subscribed or at least have search for a community outside of your instance before you can interact with it within your instance.
Actually, no. If at least one user from your instance has subscribed to a community of another instance, that other instance’s communities show up in your search results (they are federated)… at least that is how it should work 100% of the time, but it doesn’t. In most cases, it doesn’t actually show all of the communities that would show up during a search, meaning the web UI would return a lot more results.
The method you’re using is actually a safe bet: specify the instace’s URL and don’t have to rely on whether that specific instance is federated with my instance, just search for communities there as well. That approach is easier if you actually have a link to the community which you’d like to add and the instance on which it resides. But, if you have no idea on what instance that community might reside, it’s easier to actually seach for it via the web UI, which I hoped would be implemented in an app sooner or later (having relevant results I mean).
Yeah I’m wanting to implement that into liftoff. That’s the main feature I’ve been wanting in apps.
To clarify this app is a resurrection of an abandoned Lemmy app, when this app was active Lemmy was not as popular and we didn’t have massive instances like Lemmy world. Working through this is a fun challenge but it’s also a user experience challenge. What instance are you having issues with?
This seems like a golden opportunity for distros like Suse and Ubuntu, who offer enterprise support for their free product, to poach some RHEL customers.
I wish Ol’ Debian would get the love it deserves, especially for enterprise where their “stability over the latest flashiest software” philosophy should really shine. People on the desktop side criticizing how slowly Debian packages update is generally responded with “well it’s a server OS first and foremost, the Debian derivatives are more suited for desktop,” so why does no one use Debian for servers? And as far as I know Debian has always prioritized stability and reliability above anything else, and have never pulled any sort of corporate antics even close to what Canonical and Red Hat have pulled.
I’m using it. Almost 200 servers at work. No problems whatsoever. I almost smile reading news like this, because it shows me I did the right thing betting on debian
I literally just got Rocky installed on two servers. We were going to field test before migrating to a paid subscription. That sure as hell isn’t happening now. If IBM cannot help but to bite the hand that fed it, then I have little confidence they aren’t going to turn into another Oracle.
If IBM cannot help but to bite the hand that fed it
Which hand was feeding them and how are they biting it?
Idc about IBM but the saying makes little sense to me here. It’s not like Ubuntu annoying Debian which at one point was their upstream. They find upstream up to fedora and beyond.
If anything, Distros like Alma and Rocky bite the hand that feeds them by offering paid support contracts. Nothing is illegal about that. But I think the saying fits the reverse better.
linux
Hot
This magazine is from a federated server and may be incomplete. Browse more on the original instance.