No, I didn't! But I was happy to see such a makefile target exists and one doesn't need to go through console logs to see wtf was installed, or just shrug and think "yet another project made like it's the center of the universe"
It's is pretty cool, I can feel the energy poured into making your personal computing experience more seamless! Though the first thing I wondered when reading your examples is "how is ambiguity resolved?". Like albums, artists and tracks having the same string or sharing a prefix (search in this specific order, I guess?); or artists having the exact same name.
The aspect I like the most is using the filesystem as a database, since that's what UNIX people should like (and you can use symlinks for more complex cases). In fact, I myself made a music player with that as central philosophy, though it is much more bare/suckless compared to yours: https://git.sr.ht/~q3cpma/mus
Did you consider implementing a simple event system (maybe even IPC) for track and status change? Possibly MPRIS or something simpler. That was the main feature I kept from cmus when creating mus, so that I can easily interact with it through lemonbar and scripts.
Thank you. To answer your first question, ambiguity isn't resolved unfortunately. When the album has the same name as the artist for instance, I have sometimes resorted to renaming the album name by adding "album" to it. You can however get an exact search by adding -e so that resolves some problems.
I love the idea. My music is now 50% cloud only, 50% on disk. I mean, its 100% in the cloud, i just have local files for half available. Ive been thinking about self hosting some music provider thingy (or even just supporting ssh via my dyndns-like capability) to my NAS and bringing music back to self owned files. However, it is work to do when the internet is pretty reliable, costs are low, etc.
Those who love this conceptually but have/had cloud music, did you act? How/why?
I used various Subsonic clients for a number of years, but the clients were always lacking. Android clients were buggy or didn't prioritize local caching and I preferred to use mpd+ncmpcpp on my laptop.
I ended up switching to fully-local media after realizing that my 956GB flac+mp3 would be ~159GB when converted to Opus. I now use https://github.com/nvllsvm/harmonize to maintain a 128kbps Opus version of my main library and Syncthing to synchronize it to my phone and laptop.
---
side note, Auxio is the client I'm using on Android with my synced library.
I host my music on an Unraid box on my home LAN, use Plex+Plexamp to serve and access it, and 1000% recommended Plex+Plexamp. It is, without a doubt, the best music app I've used on a phone, including other self-hosting solutions, local apps like Poweramp, and commercial streaming apps like Spotify. I really can't speak of it highly enough.
I've also been a long time airsonic (and now airsonic-advanced) user for so long I can't even remember, but a couple years ago I switched to navidrome which is also subsonic compatible and it's sooo much nicer.
It doesn't fit my use-case very well, though. I'm not saying it needs to, but I'm going to put my use-case out there in case someone is looking for project ideas.
We have oodles of music players on Linux, GUI and terminal. But we have very few choices that
* are optimized for the absurdly, comically large library of someone who has been diligently collecting and organizing music for decades
* collect playback statistics and allow user rating of songs
* that can be used to create smart playlists
I used amarok for years, but it keeps dying and reviving, and I don't trust it to stick around. I then used mpd for years, but while mpd excels at large libraries, the other two requirements have to be implemented client-side, and the experience was always at least a little janky. I currently use Strawberry, but 1) it chugs with a large library, 2) its smart playlists aren't expressive enough, and 3) it is also kind of janky, and I experience frequent crashes.
The only player I've found that really fits my use-case like a glove is MediaMonkey, but I walked away from Microsoft years ago, and I'm not about to go back now just to wrangle my music library.
Nice, thanks for sharing your thoughts. Currently, I use navidrome[1], which not really is a player but more a music server, but since it supports the "subsonic" protocol, you can use native apps to connect and manage your stuff (substreamer for android / iOS is all I really need but navidrome also comes with a handy web interface). It also has support for json based smart playlists[2].
Just to add an alternative, I'm using Airsonic Advanced[0] as my subsonic server of choice if for only one reason: it properly supports folder navigation. I've ranted about this before (looking at you Jellyfin) but my folder layout is sacred and any media service I use needs to respect it.
For an android client I use tempo[1] which again was one I landed on because pretty much all the other clients didn't support folder lookup either (I think dsub also does but tempo is a lot prettier).
That looks like it checks most of my boxes, but I have a personal/philosophical objection to running a service. The objection is, I don't want to[0]. I just want a local application. Not local-first, I want local-only. Just an application.
[0] and also I think it's insane to add that much complexity to something that is single-user.
I'm also a navidrome user and I run it via docker exposed via traefik so I can access my music anywhere. I can use any subsonic client on android or iOS and I can bluetooth that to my car or headphones or whatever and I can load it up on my laptop anywhere.
As you've said you just want a local application just wanted to mention that in case that's actually something that might also be useful for you.
Totally understandable. I recently thought of developing a cross platform player in C# and AvaloniaUI, but cross platform audio is not as easy as it seems, especially trying to use open source libs only and minimizing dependencies.
Yes, while a comically large music library is supported in principle (kew offers to cache your library if it takes a long time to search through), it might not be entirely suited for it.
As for your other two suggestions those fall outside the scope of kew. kew is supposed to be simple with minimal bloat.
Strawberry is a pretty solid Amarok fork that is picking up steam. They are now releasing multiple releases a month and in my opinion it's a great "fully featured", gui first, easy to use player that handles large libraries well.
I've been using Plexamp for the past 4 years, and its been great for music on my phone and computer. One thing it lacks is a good offline mode. Recently I started syncing my media to my laptop as a backup, and as a way to locally play 5.1 FLAC audio (Plexamp doesn't support 5.1 audio).
I'll have to check this out. I wonder if it can play 5.1 audio correctly?
Defaulting to the only result in your library is perfectly reasonable, but it the case of a collision, what does it do?
My guess is that it runs through in a loop (I haven't looked at the code) and simply goes with the first result that it finds, but that doesn't cover all possible use cases.
kew dir <album name> (sometimes it's necessary to specify it's a directory you want)
kew song <song> (or a song)
kew list <playlist> (or a playlist)
The directory can be an artist or an album, so there's still ambiguity there. But kew cannot differentiate between the two. It matches against files or against directories.
You know I'm not sure. I think there was some problem with FFmpeg in earlier versions, but I'm not 100% sure. There shouldn't be a problem trying to install it in earlier versions though.
So much thanks for giving a good example
I agree it's important. kew is so small it was pretty trivial to do.
The aspect I like the most is using the filesystem as a database, since that's what UNIX people should like (and you can use symlinks for more complex cases). In fact, I myself made a music player with that as central philosophy, though it is much more bare/suckless compared to yours: https://git.sr.ht/~q3cpma/mus
Did you consider implementing a simple event system (maybe even IPC) for track and status change? Possibly MPRIS or something simpler. That was the main feature I kept from cmus when creating mus, so that I can easily interact with it through lemonbar and scripts.
Yes, MPRIS is supported.
Those who love this conceptually but have/had cloud music, did you act? How/why?
I ended up switching to fully-local media after realizing that my 956GB flac+mp3 would be ~159GB when converted to Opus. I now use https://github.com/nvllsvm/harmonize to maintain a 128kbps Opus version of my main library and Syncthing to synchronize it to my phone and laptop.
--- side note, Auxio is the client I'm using on Android with my synced library.
It works phenomenally.
At some point I was going to mirror it locally, but never got around to it.
It is all backed up in dropbox
Use whatever you want! Just wanted to suggest it.
It doesn't fit my use-case very well, though. I'm not saying it needs to, but I'm going to put my use-case out there in case someone is looking for project ideas.
We have oodles of music players on Linux, GUI and terminal. But we have very few choices that
* are optimized for the absurdly, comically large library of someone who has been diligently collecting and organizing music for decades
* collect playback statistics and allow user rating of songs
* that can be used to create smart playlists
I used amarok for years, but it keeps dying and reviving, and I don't trust it to stick around. I then used mpd for years, but while mpd excels at large libraries, the other two requirements have to be implemented client-side, and the experience was always at least a little janky. I currently use Strawberry, but 1) it chugs with a large library, 2) its smart playlists aren't expressive enough, and 3) it is also kind of janky, and I experience frequent crashes.
The only player I've found that really fits my use-case like a glove is MediaMonkey, but I walked away from Microsoft years ago, and I'm not about to go back now just to wrangle my music library.
1: https://www.navidrome.org/ 2: https://github.com/navidrome/navidrome/issues/1417
For an android client I use tempo[1] which again was one I landed on because pretty much all the other clients didn't support folder lookup either (I think dsub also does but tempo is a lot prettier).
0: https://github.com/kagemomiji/airsonic-advanced
1: https://github.com/CappielloAntonio/tempo
[0] and also I think it's insane to add that much complexity to something that is single-user.
As you've said you just want a local application just wanted to mention that in case that's actually something that might also be useful for you.
As for your other two suggestions those fall outside the scope of kew. kew is supposed to be simple with minimal bloat.
Is this a nix thing (i'm unsure what freeimage-unstable is)
The version of kew packaged for Nix is very old: v1.5.2. We're at version 2.8.2. So it's more than a year old, from very early on in the project.
I don't know how relevant these vulnerabilities are to kew, which isn't run across the network in any way, it just reads your local files.
Thank you for bringing this to light. I don't know how feasible it is to use something other than freeimage though, gonna have to investigate.
Looks like a nice project, I like the terminal album art display :).
https://www.musicpd.org/
I'll have to check this out. I wonder if it can play 5.1 audio correctly?
[0] - https://musikcube.com/
And I "compiled from source" as I am using Fedora, but it was just one command.
Thank you!
This here is even better, I love the minimal approach.
Defaulting to the only result in your library is perfectly reasonable, but it the case of a collision, what does it do?
My guess is that it runs through in a loop (I haven't looked at the code) and simply goes with the first result that it finds, but that doesn't cover all possible use cases.
Unlike the -e solution, it wouldn't make the command significantly more verbose than the default option.
You could also print a list of commands for the specific options (or allow for index based selection) in cases where there were collisions.
kew dir <album name> (sometimes it's necessary to specify it's a directory you want)
kew song <song> (or a song)
kew list <playlist> (or a playlist)
The directory can be an artist or an album, so there's still ambiguity there. But kew cannot differentiate between the two. It matches against files or against directories.
% brew install kew
...
kew: Linux is required for this software.
Error: kew: An unsatisfied requirement failed this build.
EDIT: Added.