utsushi: 3.57.0 -> 3.59.2

[wucke13]

Motivation for this change

• adding installChecks
• building from gitlab source now

Things done

• Tested using sandboxing (nix.useSandbox on NixOS, or option sandbox in nix.conf on non-NixOS)
• Built on platform(s)
  • NixOS
  • macOS
  • other Linux distributions
• Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
• Tested compilation of all pkgs that depend on this change using nix-shell -p nix-review --run "nix-review wip"
• Tested execution of all binary files (usually in ./result/bin/)
• Determined the impact on package closure size (by running nix path-info -S before and after)
• Ensured that relevant documentation is up to date
• Fits CONTRIBUTING.md.

Notify maintainers

cc @

[symphorien]

Any reason to remove the patches which make this work without imagemagick in the PATH ? See #64985 (comment) for motivation.
cc @thedavidmeister for testing.
Nice to see that you enabled the tests.

[wucke13]

@symphorien no good ones. I just assumed they were some magic for making the epson distributed thing tarball work without actually looking into them. I'm sorry. I will incorporate them again and test the build, though I prefer my way of wrapping tesseract. What do you think in that regard?
The manual states wrapping should be avoided in favor of patching in the configurePhase (https://nixos.org/nixpkgs/manual/#ssec-stdenv-functions), so the patches seem to be the best approach!
Apparently, the patches do break the instalCheck. I will investigate further.
I removed the patches, and wrapped the files in question. If somebody has one of the compatible scanners, please try this out!

Using 597c8058fec8faa766e8ddfc64cb9370f4dc5e4f I was able to scan using the utsushi sane backend, but not using the utsushi binary (both only tested via usb).

[symphorien]

Patching is required for imagemagick because otherwise all programes using libsane should be wrapped as well, which is not possible. So please revert to using patches.

[wucke13]

@symphorien I do not understand about what libsane are we talking? Like what breaks due to libsane being wrapped?

[symphorien]

If you look at the imagemagick patch, you will see that the C code calls convert explicitly. And any program on the system can load libsane and thus call this code. Therefore any program can require convert on PATH. The only reasonable solution is to replace convert by an absolute path in the C file.

[wucke13]

@symphorien Ok, done. This should cover the get-text-orientation script too.
WIth df789f2 both utsushi and xsane work for scanning (via usb).

[thedavidmeister]

@wucke13 ok i'll try testing then

[thedavidmeister]

xsane fails to scan:

[thedavidmeister]

similar for utsushi, the scanner runs but files aren't created

[wucke13]

@thedavidmeister please give me detailed information about the setup on which you're having trouble scanning. As stated, I was able to scan both with utsushi and and xsane, so your feedback surprises me a little.

I did it like this
configuration.nix:

{
  nixpkgs.config = {
    packageOverrides = super: let self = super.pkgs; in {
      current = (import /* insert absolute path to checkout of my branch, ending with `/nixpkgs` */ {
        config = config.nixpkgs.config;
      });
    };
  };
# ...
  hardware.sane = {
    enable = true;
    extraBackends = with pkgs; [ current.utsushi ];
  };
# ...
  environment.systemPackages = with pkgs; [ xsane ];
}

The scanner that I was using is the integrated scanner of an Epson ET-3750.
So, how what exactly did you do what ultimately failed?

Edit:
Using 69b5239 I'm still able to scan both using utsushi and xsane. Please note: I have to disconnect the USB cable when switching from one software to the other, otherwise the scanner is not recognized by the other software. Furthermore, after the rebuild, a logout is necessary for sane to work, as sane uses Environment variables to specify which configuration to use, and those happen to be only reloaded on login.

[wucke13]

ping @thedavidmeister

Notice: It is somewhat pervert having something automatically merge broken = true commits on a package causing a merge conflict for a PR on that very same package where the PR potentially solves the brokenness.

[grahamc]

@GrahamcOfBorg build utsushi

[grahamc]

(note it wasn't a bot, but #70645 and #68361)

[grahamc]

@GrahamcOfBorg eval

[wucke13]

@grahamc Please backport too, I would like to not wait another 5 1/2 months until I can scan with my scanner on a stable system... Also: thank you for taking the time on this matter!

[grahamc]

@wucke13 Thank you! Please open another PR backporting ot 19.09, with git cherry-pick -x 9eb00acbbee63d20a6ac7c124005089a8bebcfba.

[thedavidmeister]

my testing was pretty simple, i set my nixos channel to the fork commit and ran nixos-rebuild

i have all the packages included as explained above, i was using utsushi when it was first merged

as this is merged i'll swap back to the latest master, rebuild nixos and see what happens 🤷‍♂️