Nvidia provides good support for CUDA on Linux, but not good support for graphics on Linux. You recognize this, and use a different brand GPU for graphics. The Nvidia users that Sir_Cmpwn is frustrated with are those that think that the same GPU should be good for both tasks, and that everyone working on graphics do all the extra work to adapt to the whims of Nvidia engineers who don't care about their graphics use-cases.

The comment that Sir_Cmpwn was replying to litterally said "Nvidia is absolutely the only option, and it is everything else that has to adapt to it." Now that is the harmful attitude. Because you use Nvidia for CUDA doesn't mean that you need to use Nvidia for graphics. If you do choose to use Nvidia for graphics (despite a clear compatibility warning from the community), you shouldn't demand the graphics people to bend-over-backwards and "[have] to adapt to [Nvidia]."

And you, coffeecat, don't seem to demand that.