On: Nvidia Removed Linux Driver Feature Due to Windows

[pyre]

That may be true, but they are presumably already maintaining a bunch of #ifdef's for the Windows-only features.

[lilyball → pyre]

Given the relative popularity of Windows vs Linux, it probably is worth the effort to maintain Windows-only features.

[67726e → pyre]

True, but Windows has the much larger market share so it becomes worth it.

[chris_wot → pyre]

> That may be true

We'll never know. That's a binary blob on your machine.

[pyre → chris_wot]

This is true, but I fail to see the relevance here. This post just comes across as a "any chance to mention Free Software" type post. This thread is talking about if this decision by Nvidia makes sense. Trying to inject "well, we could all just ignore Nvidia if you were using Free Software," doesn't really contribute much to that conversation.

> That's a binary blob on your machine.

On my machine? It's less likely than you might think.

[kbenson → pyre]

Being able to assume the #ifdef for features go one direction can have some benefit in reasoning about the code it relates to. In other words, there may be benefits to keeping the windows features a superset of the linux features, most likely in preventing stupid programmer errors.

[subway → pyre]

It goes (or at least at one time went) well beyond that. Last I heard, nvidia.o was shared across Linux, FreeBSD, Solaris, and Windows.

[ErsatzVerkehr → subway]

Don't these systems use different object-file formats?

[subway → ErsatzVerkehr]

At the very least I can confirm the object is identical (same md5sum) across Linux, FreeBSD, and Solaris. Past sources have indicated to me Windows shares his object.