speechd: introduce minimal variant, use it everywhere, inject big variant impurely

[K900]

Description of changes

TODO: figure out how to default this to false without breaking people.

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 24.11 Release Notes (or backporting 23.11 and 24.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[mweinelt]

Thanks, i think that looks worth a try.

[vcunat]

Now that I think of it, I'm not sure if services.speechd is the best place, maybe programs.speechd?

I don't care really. But now when it's new (not even on master yet) such changes are easiest. So I'm just posting this note...

[K900]

It is conceptually a service, it's just started weird currently. The upcoming next release will be a proper systemd socket activated thing.

[vcunat]

Oh, OK. BTW, another way of saving space would be to default to English only (and perhaps just a single voice) unless you override your i18n.supportedLocales or something. That looks like ~7 MB only.

[NobbZ]

Something in the changelog would have been nice… Also what are the consequences of force disabling it? I used my system for severel years without it, and have no clue how I had to trigger it right now…

[K900]

It's perfectly fine to disable it if you don't need text to speech.

[NobbZ]

And why did it get enabled by default without a mention in the changelog?

[tejing1]

Which more than 99% of people don't... so why is it enabled by default for everybody?

[K900]

Because accessibility is important. It should have definitely had a release note, though.

[NobbZ]

Yes, accessibility is important, I agree, though so far there is some random envvar set without any documentation of how to use it nor what programs can actually use it.

The closure doesn't seem to be large in its effect on my system, I still consider the way it gets introduced worth questioning.

[K900]

That environment variable gets picked up by libspeechd in applications. The whole thing was actually an attempt to reduce closure size - previously you'd just get speechd as a single package, including the gigabytes of voice data; now individual applications only depend on the library, and the daemon and voice data are separate and late bound, so you can disable them.

[tejing1]

So it's enabled for everybody because otherwise people who used it the way it worked before would have it break?

[K900]

Indeed, and also because people that rely on screenreaders should be able to use the stock configuration on the installer and on the default system.

[SuperSandro2000]

You get rid of it with:

services.orca.enable = false;
services.speechd.enable = false;

[rnhmjoj]

I think it would have been better to introduce an accessibility top-level option to centralize the default configuration of programs like brltty, orca or speechd. This is yet another undocumented default that's hard to track down.