@stigtsp, @zakame, any concerns here?

@zakame, the additions of meta.mainProgram definitions were done using a script I created to identify packages in nixpkgs that provide a single executable in bin/ where the name of that executable differs from the package's name or pname, and then automatically insert the meta.mainProgram definition for those packages.

I used the script to generate the changes in this PR, then I manually reviewed the changes focusing on cases where the executable's name in the added meta.mainProgram was surprising to me given the package's name or description. For those cases I researched the executables more, and removed the meta.mainProgram definitions if it really didn't make any sense to me for them to be blessed as the mainProgram (as well as added them to a list of packages to exclude in the future). While reviewing the changes I also did a little cleanup as mentioned in the PR's description, and I did spot check a few of the new meta.mainProgram definitions by running nix run.

Just mentioning all that in case you were concerned that I performed all these edits manually, which would be a lot more error prone.

My goal, in creating this script, and using it to generate this PR (and past PRs, e.g., #172949, which collectively have added meta.mainProgram definitions to a few hundred packages now), is to get nix run working with as many packages as possible in nixpkgs. It's definitely possible that this process could result in a few erroneous additions of meta.mainProgram definitions. However, given that the cost of such errors is very minor (e.g., nix run failing to find an executable that matches what was defined for the packages mainProgram), and that manually running nix run for each package (168 packages for this PR) is very time consuming, I don't think the tradeoff is worth it.

@SuperSandro2000 or @zakame, just a polite poke in case this has fallen through the cracks.

@SuperSandro2000, anything I can do here to get this merged?

I am not sure. I would appreciate @stigtsp 's feedback.

@malob see also #176398 from @zimbatm

also not familiar with implications of adding mainProgram

Just meta information for nix run.

Are there tests that check if mainProgram resolves/executes somewhere?

nope

Alrighty I've,

• rebased on master,
• removed redundant mainProgram definitions due to perlPackages: add default meta.mainProgram #176398, and
• moved all indentation changes to a 3rd commit which fixes indentation issues in the whole file.

@stigtsp, sorry if the indentation changes and the reshuffling of attributes made the commits harder to review. I have (what I think is generally a good) habit of making small fixes/improvements unrelated to the changes I want to make when I notice opportunities to do so. In many cases these changes are minor enough that they don't warrant a separate commit, or like in the case of reshuffling the attributes, I wouldn't bother to make those improvements if I had to put them all into a separate commit (since it's annoying to note them all down, and come back to them separately).

So for changes like the shuffling of attributes, I usually lean towards including them, even if it makes the commit a little harder to review, since the alternative of in my case would be to just leave those changes off (which seems worse, at least to me, overall).

However, I agree with you that the indentation changes in particular definitely deserved a separate commit since the indentation errors are easy to find on correct on their own, and they made reviewing the other changes a lot more annoying.

Politely poking @stigtsp and/or @SuperSandro2000