> Another way is to eliminate the dependency

That'll resolve the violation for future releases. However, all previous releases are still infringing.

For a violating company who really doesn't want to open source their project, their best bet would probably be to (remove the dependency and) pay damages for previous infringement.

You'd hope damages in a case like this would be small given it went unnoticed for so long. Considering the shared-mime-info project itself is not commercial software, there probably wasn't significant damage to the project or the authors.

> you may still have to pay damages

This is probably a first time I've seen damages mentioned in relation to GPL violations. Did anyone try enforce this?

IIRC, there was someone who had written some networking code in Linux and independently started sueing hardware vendors for GPL violations. The Software Freedom Conservancy said that he was doing more harm than good, and said that if someone is violating the GPL, lawsuits should first only require compliance with the GPL, then seek punitive damages if they fail to comply.

There's a site here https://gpl-violations.org/news/ which has some cases where there have been legal actions related to GPL violations.

There is at least one case[0] I can find. Probably it is exceedingly rare simply because companies are much more likely to settle, especially in the cheapest way possible i.e. stop distributing the tainted software.

[0]:https://wiki.fsfe.org/Migrated/GPL%20Enforcement%20Cases#Bus...

To be clear, stop distributing is often "good enough" but technically damages could still be sought for copyright violation.

It's pretty clear that distributing without a license (or in violation of one) is copyright infringement, and that's subject to damages.

However, most non egrigious copyright infringement cases are more about stopping future infringement than damages. So I'd be surprised to see much GPL enforcement with damages.