Essentially, an updated dependency requirement in Mesa (updated Zlib) broke an important benchmarking tool (SPECViewPerf) used by hardware vendors. Subsequently, this change was reverted. This caused a debate in the Mesa dev community, with some devs claiming it’s not Mesa’s fault, it should be treated as a bug in SPECViewPerf instead. In response, AMD’s Mesa dev said this isn’t a technical issue, but rather a political/strategic issue - you don’t want to anger important workstation vendors and other high-level parties who use this tool, especially since they contribute so much to the Linux ecosystem. That would make the Mesa project seem very immature/unreliable.
As an example, imagine if this change broke something more popular like Steam - Valve and all Linux gamers would be out for blood and you bet the Mesa change would be reverted without debate - even if they were technically in the right (that it’s not a bug).
So this incident serves as an important reminder for those who work on big opensource projects like this - just because your actions are technically correct, it doesn’t mean it’s okay to break everyone else’s stuff, expecting they’ll fix it. This is in fact something Linus preaches when it comes to kernel dev - “don’t break userspace”.