Consider using either homebrew ignore/detach or pin for casks

[GhostLyrics]

Yes, I have read Homebrew/homebrew-cask#44914 and #4898 but none of those provided an example of breaking existing workflows.

A detailed description of the proposed feature

Allow some formulae that were installed to be ignored by brew upgrade - either by pinning or by completely making homebrew forget they were installed via its system.

The motivation for the feature

I'm currently on v12 of navicat-for-mysql. v15 which has at some point become the default for homebrew will require my company to purchase a new license. I don't have plans to upgrade currently, but would still like to use brew upgrade (without arguments) for everything else I have installed.

With the new behaviour of brew upgrade now also upgrading casks, brew upgrade will always do the wrong thing for me - upgrade to a new major and essentially force a purchase.

How the feature would be relevant to at least 90% of Homebrew users

Avoid upgrading a commercial cask to a new major version that requires a new, potentially unavailable license.

What alternatives to the feature have been considered

• Detaching the app from homebrew inventory completely after initial installation

[claui]

@GhostLyrics If there’s no navicat-for-mysql12 cask available on homebrew-cask-versions, you can create a custom tap and copy/paste that cask there (it’s documented and takes a few minutes) and you should be good to go.

In fact, migrating individual casks to a custom tap is straightforward enough that adding any pinning/detaching feature wouldn’t justify the development and maintenance resources needed for the feature.

Closing for now but feel free to comment for further discussion.

[GhostLyrics]

I created my own custom tap to solve my problem. It took me a while to find out that extract does not work with casks, which I had previously used to create a custom (local) tap for a different issue.

Thank you for your hints.
Do you think it's worth submitting that as a PR to homebrew-cask-versions?

[claui]

Thank you for your hints.

You’re welcome @GhostLyrics!

Do you think it's worth submitting that as a PR to homebrew-cask-versions?

I’d say it’s not. Looking at cask analytics, even the latest version has less than two installs per day. Having a N − 2 version on cask-versions is probably not useful enough.

[gromgit]

I created my own custom tap to solve my problem.

Side note: Custom taps (at this stage of Homebrew evolution, anyway) need not be managed by Git at all, much less require remote hosting. I personally did a brew tap-new gromgit/test some months back as a formula testbed tap. It resides only on my MacBook's SSD (and my backup drive), and it doesn't have a .git subdirectory. Homebrew still recognizes it as a tap, since it's in the right place with the right contents, but simply skips over it when it comes time to brew update.

So if anyone thinks "aw hell, I gotta maintain my own repos just for one tap?!?!", think again. Taps are ridiculously easy to deal with.