Linus Torvalds Questions The Not So Glorious Driver For That Funky Looking RGB Mouse

[andyprough]

I bought a marginally good (at best) Corsair Dark Core RGB a few weeks ago and it is completely gimped outside of basic usage on Linux. As a side note its a generally horrible mouse with crappy battery life and the corsair iQue software is terrible too on windows!

All that said it would be nice to have it workable on Linux. I just this morning bought a Razer Naga to replace this otherwise new mouse because the Naga apparently is much better supported.

a) why did you buy it? b) why should the kernel support it? c) why haven't you returned it?

Not judging, just wondering. I've made a lot of purchases like that myself. 10 years later I still have trouble throwing it out, even though it will never work again on any machines I will own.

[guus]

Arguably, it would ideally be the other way around. If Windows was stricter about not supporting any device that doesn't conform to the standard, the device would actually have been designed to conform. Then any driver that follows the standard would support it properly, without requiring arcane study of exactly in what manner different devices violate the standard.

I'm not too familiar with the USB HID standard, or exactly how different devices violate it. But if the standard is sensible, it's much better for everyone to strictly conform to it.

Arguably, device makers should follow the specs. But life isn't perfect, mistakes are made. As Jon Postel said: "Be liberal in what you accept, and conservative in what you send." This helps interoperability (both for computers and for humans amongst each other). From a security standpoint, you don't want to be too liberal though. But there are cases where you could easily allow violations of the specifications while still providing some sensible behavior without any security risks. So the core HID driver could be written to work around some common non-spec cases, and only where these workarounds fail would we then need to have special drivers like hid-corsair.

[zexelon]

andyprough
a) Good question, to be honest my experience with Linux over the last couple of years has been extremely positive. There have been a few issues, but pretty minor. Also I didnt do my research first...
b) Must admit, I am not certain if the issue is kernel HID drivers or something that can be user spaced... not an expert in this area at all! Why does the kernel have to support the abomination discussed in the OP?
c) I have a severe aversion to returning things... I consider it a personal failure to properly do my due diligence as a buyer... yah its odd.

[edwaleni]

The question on the concept of Linux being "too precise" wouldn't be new.

I have had certain peripherals hang on Linux but work fine on Windows. In one case I found that the Linux driver had the device do exactly what it was specified and it ran really fast until it hung. Looking at the Windows driver it suppressed the specification and it ran at about 88-90% and it never hung.

[angrypie]

Arguably, device makers should follow the specs.

If they followed specs there would be no reason to ship bloated proprietary crapware along with the devices they make. It's intentional.

[zexelon]

If they followed specs there would be no reason to ship bloated proprietary crapware along with the devices they make. It's intentional.

Pfffff lets be honest specifications are just in place to stifle "creativity"... and if everything followed spec and just worked think of how many programmers would be out of jobs??? You have to think of the people!

*sarcasm*

[Viki Ai]

Pfffff lets be honest specifications are just in place to stifle "creativity"... and if everything followed spec and just worked think of how many programmers would be out of jobs??? You have to think of the people!

*sarcasm*

"Move fast and break everything. And run off with the money before anyone realises." It is the motto of the 21st century.

[tildearrow]

Effectively we have this: https://xkcd.com/927/

Avoidable with successful EEE.

[Delgarde]

Arguably, device makers should follow the specs. But life isn't perfect, mistakes are made. As Jon Postel said: "Be liberal in what you accept, and conservative in what you send."

It's a nice principle, but there's another one that also fits — garbage in, garbage out. Tolerating minor deviation from the spec is fair, but there's only so far you can go with that.

An interesting example also in my news feed today — Twitter discovering that on Firefox, private data was unexpectedly being cached. And this wasn't the fault of Firefox for caching data — it was the fault of Twitter not actually passing "no-cache" headers, and relying on browser-dependent heuristics that happened to work for them on Chrome.

[timrichardson]

Linus's strongest word when he saw something he didn't like was "hmm?".