When they do it by screwing over their users, yes.
> Introducing Linuxisms in ... Linux?
Introducing non-portable ways of doing things that were previously portable to non-Linux OSs, yes. It's certainly a minor point since if anything some of those changes increased portability within the Linux ecosystem, but I can sorta see "introducing Linuxisms in areas that previously weren't Linux-specific" (ex. `service foo start` did the same thing on RHEL and FreeBSD; `systemctl start foo` isn't going to be a thing on any BSD).
> When they do it by screwing over their users, yes.
I agree that screwing over your users is a bad thing, but I do think RH contribute a lot of time, money and effort in a way that benefits a larger Linux user base. Just by looking at Linux kernel contributions, RH are one of the largest companies contributing developer time.
> Introducing non-portable ways of doing things that were previously portable to non-Linux OSs.
I think SysVinit was going to be replaced sooner than later. There's a lot to be desired in a way that some systemd developers evangelized it, but the other major alternative (upstart) was also a Linuxism.
Even BSD systems are moving (or considering moving) away from it. From all the "standard" Unixisims, I think that's one of the parts that's aged the worst, and it was a matter of time before various operating systems (BSDs, Linux distros,...) replaced it.
When they do it by screwing over their users, yes.
> Introducing Linuxisms in ... Linux?
Introducing non-portable ways of doing things that were previously portable to non-Linux OSs, yes. It's certainly a minor point since if anything some of those changes increased portability within the Linux ecosystem, but I can sorta see "introducing Linuxisms in areas that previously weren't Linux-specific" (ex. `service foo start` did the same thing on RHEL and FreeBSD; `systemctl start foo` isn't going to be a thing on any BSD).