fix(controllers): restore legacy passing of HID data in plain array#11470
Conversation
|
I don't have a HID controller to test, so it would be nice if someone with the affected hardware could fix it.
|
|
I fully agree, that this is the right place to fix this! Thank you for providing this fix! |
|
Thank you. I don't think so. from what I can tell, these code parts are not particularly hot. Still, I'd really prefer if someone could test this before merging. |
Fixes mixxxdj#11461, alternative to mixxxdj#11463
|
confirm commit works against last pull on 2.4 for a NITrakKontrolS2MK3 tyvm! |
|
Thank you for testing. I'm still waiting on a review by @JoergAtGithub or someone else. Maybe @ywwg? Thanks. |
|
I will test it tonight! |
|
friendly ping @JoergAtGithub. If you don't have the time or resources to test this, you could probably also just merge this since it was already confirmed to be working. |
|
I will check this out this weekend - promised! |
| * @param {number[] | ArrayBuffer | TypedArray} bufferLike Object that can be represented as a sequence of bytes | ||
| * @returns {DataView} dataview over the bufferLike object | ||
| */ | ||
| const createDataView = function(bufferLike) { | ||
| return new DataView((() => { | ||
| if (Array.isArray(bufferLike)) { | ||
| return new Uint8ClampedArray(bufferLike).buffer; | ||
| } else if (ArrayBuffer.isView(bufferLike)) { | ||
| return bufferLike.buffer; | ||
| } else { | ||
| return bufferLike; | ||
| } | ||
| })()); | ||
| }; |
There was a problem hiding this comment.
| * @param {number[] | ArrayBuffer | TypedArray} bufferLike Object that can be represented as a sequence of bytes | |
| * @returns {DataView} dataview over the bufferLike object | |
| */ | |
| const createDataView = function(bufferLike) { | |
| return new DataView((() => { | |
| if (Array.isArray(bufferLike)) { | |
| return new Uint8ClampedArray(bufferLike).buffer; | |
| } else if (ArrayBuffer.isView(bufferLike)) { | |
| return bufferLike.buffer; | |
| } else { | |
| return bufferLike; | |
| } | |
| })()); | |
| }; | |
| * @param {Uint8Array | number[] | ArrayBuffer} bufferLike Object that can be represented as a sequence of bytes | |
| * @returns {DataView} dataview over the bufferLike object | |
| */ | |
| const createDataView = function(bufferLike) { | |
| if (bufferLike.constructor === Uint8Array) { | |
| return new DataView(bufferLike.buffer); | |
| } else if (Array.isArray(bufferLike)) { | |
| return new DataView(new Uint8ClampedArray(bufferLike).buffer); | |
| } else { | |
| return new DataView(bufferLike); | |
| } | |
| }; |
There was a problem hiding this comment.
I noticed a performance impact by your implementation. I played a bit with the code and this is faster. I guess because the bufferLike is in nearly all cases a Uint8Array.
I don't think there is a need to support any other TypedArray type.
There was a problem hiding this comment.
I don't think there is a need to support any other TypedArray type.
I disagree, and I don't think it would be difficult to support it.
The improvement from performance is likely just the change in ordering of the if cases, can you experiement whether there is a significant difference when you change bufferLike.constructor === Uint8Array back to ArrayBuffer.isView(bufferLike) in your suggestion? Also I doubt the IIFE is make any difference either because it should just be inlined by the JIT.
There was a problem hiding this comment.
Thee IIFE is removed for better code readabilty and I also doubt it affects the performance. I will benchmark your suggestions.
There was a problem hiding this comment.
Thee IIFE is removed for better code readabilty
I guess this is certainly something to argue about, but I think the IIFE improves readability because it DRY's up the code. If I see the same code repeated in all branches, it usually creates suspicion for me that this could be the source of a bug from mindlessly copy-pasting code. But that is certainly controversial.
|
I tested this under various conditions on Windows 7 and Windows 11 and it always worked. There is a slight performance degrations but this is not so clear to measure. Therefore this is Ok for me, further performance improvements can be done in Follow-Up PRs if needed. |
|
Thank you for the review. 👍 |
3 participants
Fixes #11461, alternative to #11463