copyparty depends on vips which brings lots of dependencies

[kidonng]

Output of brew config

HOMEBREW_VERSION: 5.0.12
ORIGIN: https://github.com/Homebrew/brew
HEAD: d01011cac6d72032c75fd2cd9489909e95d9faf2
Last commit: 5 days ago
Branch: stable
Core tap JSON: 30 Jan 05:09 UTC
Core cask tap JSON: 30 Jan 05:10 UTC
HOMEBREW_PREFIX: /opt/homebrew
HOMEBREW_CASK_OPTS: []
HOMEBREW_DOWNLOAD_CONCURRENCY: 28
HOMEBREW_EDITOR: code --wait
HOMEBREW_FORBID_PACKAGES_FROM_PATHS: set
HOMEBREW_MAKE_JOBS: 14
Homebrew Ruby: 3.4.8 => /opt/homebrew/Library/Homebrew/vendor/portable-ruby/3.4.8/bin/ruby
CPU: 14-core 64-bit arm_brava
Clang: 17.0.0 build 1700
Git: 2.50.1 => /Library/Developer/CommandLineTools/usr/bin/git
Curl: 8.7.1 => /usr/bin/curl
macOS: 26.2-arm64
CLT: 26.2.0.0.1.1764812424
Xcode: N/A
Rosetta 2: false

Output of brew doctor

Your system is ready to brew.

Description of issue

Sorry if this should have been an issue in homebrew-core.

Until Homebrew/homebrew-core#264919, installing copyparty via Homebrew will also install gcc (copyparty -> vips -> fftw -> gcc). A portable file server that comes with a compiler, what a suprise!

Even if fftw no longer depends on gcc now, vips is still a heavy dependency:

$ HOMEBREW_NO_ENV_HINTS=1 brew deps copyparty | wc -l
     110
$ HOMEBREW_NO_ENV_HINTS=1 brew deps vips | wc -l
     102

vips is but one of the optional dependencies of copyparty. I don't see why ffmpeg is in caveats but vips is not. Or why the lighter alternative pillow is not used (I don't think anyone using copyparty would be concerned about thumbnail generation performance).

Alternatively, in light of Homebrew/homebrew-core#261303, vips can also be slimmed down.

[kidonng]

I'm going to be a little off-topic and talk about the situation more.

While copyparty is my main concern, I would also like to know the rationale behind how dependencies are decided for formulae in general.

Yes, this is inspired by the ffmpeg-full discussions. Yes, I have read them all. I support the decision to reduce less used dependencies, especially runtime ones that can be installed and detected easily.

Take another example, mpv now comes with deno (mpv -> yt-dlp -> deno).

The latter (yt-dlp depending on deno) is fine since most use it for downloading YouTube videos (it's in the name), and that requires a JS runtime, which upstream recommends deno.

But I'm not sure about the former (mpv depending on yt-dlp). mpv don't need it for online media you can get a direct link of, only selective websites, for which you usually get a worse streaming experience (suboptimal buffering, losing thumbnails or playlists, etc.) I assume most mpv users just use it for local media.

All in all I wonder to what extent is "reducing dependencies" or "it just works" goes. ffmpeg and imagemagick are both popular so it's worth bring down the size, while copyparty and mpv are power user software so just maximize capabilities? It's this vagueness I'm mostly concerned with.

[SMillerDev]

While copyparty is my main concern, I would also like to know the rationale behind how dependencies are decided for formulae in general.

If there is demand for an addition, and people make a pull request for it they are generally added.

ffmpeg and imagemagick are both popular so it's worth bring down the size, while copyparty and mpv are power user software so just maximize capabilities?

People often complain about "why did brew install ffmpeg install so many things", I've never seen a complaint about "why did brew install copyparty install so many things". There is no way to define what software needs, because everyone has a different usecase. That's why the best we can do is just listen to the community and do what we think the majority wants.

[daeho-ro]

You can open a PR, here is the one.

• copyparty: change dependency for thumnails homebrew-core#265205