bug: add aes support for darwin/arm64 #5
Conversation
| @@ -1,4 +1,4 @@ | |||
| //go:build !android || !arm64 | |||
There was a problem hiding this comment.
this was a bug, wasn't it?
| @@ -1,4 +1,4 @@ | |||
| //go:build !android || !arm64 | |||
| //go:build arm64 && !android && !darwin | |||
There was a problem hiding this comment.
I'm unsure about the original intention, but I think this is the right thing to do here...
There was a problem hiding this comment.
I think we should discuss this in our 1:1 today. It may also be that a better pattern would be to have files for each combination of GOOS and GOARCH we support rather than doing smart things with conditionals (as you mentioned, we may already had some issues with conditionals, hence it seems it's better to be a little more humble here and avoid too many conditionals).
There was a problem hiding this comment.
We discussed this topic in our 1:1 and @ainghazal is going to write a small internal document to evaluate our options here. We'll explore possibilities such as vendoring internal/cpu.
There was a problem hiding this comment.
Once we have explored options, we'll try to figure out the cheapest and most convenient for us way to fix the bug and ensure that we're in a place where darwin/arm64 is WAI.
This diff ensures we hardcode the capabilities of darwin/arm64. While there, strive to make the packages naming slightly less confusing. Reference issue: ooni/probe#2122. Surpersedes (and draws from) #5. Co-authored-by: ain ghazal <[email protected]>
|
Thanks! I have build upon your patch here at #7. |
This diff ensures we hardcode the capabilities of darwin/arm64. While there, strive to make the packages naming slightly less confusing. Reference issue: ooni/probe#2122. Surpersedes (and draws from) #5. Co-authored-by: ain ghazal [email protected] Co-authored-by: ain ghazal <[email protected]>
x/sys/cpulacks support fordarwin/arm64, so here I am porting over the hardcoded implementation frominternal/cpu.For reference: