Feature/mac arm64 #38
Conversation
| # ranging from false positive (only copying a prefix of SkRRect from serialized form) | ||
| # to possibly useful (memcpy() with non-trivial types). Annoyingly you can't really | ||
| # break them up any finer. | ||
| # TODO(mtklein): suppress / fix each site as appropriate? |
There was a problem hiding this comment.
Who is @mtklein ? Does this code from upstream ?
There was a problem hiding this comment.
I bet that's me!
This comment and suppressed warning was in upstream Skia for a couple weeks, but removed in December 2019. You can see the timeline at https://bugs.chromium.org/p/skia/issues/detail?id=9674. I think we fixed whatever was bothering -Wclass-memaccess at the time.
There was a problem hiding this comment.
yes, this code from upstream, I just cherry pick minimal changes needed to compile in M1.
There was a problem hiding this comment.
Ah ok Github showed you were author. So you Co-authored the first commit and fully the second one ?
Does xcrun esy works ? and does our upstream mono/skia (We are fork of fork) compile fine ?
There was a problem hiding this comment.
@Et7f3 yes, that also works and it produces an arm64 library, previously it produces an x86 library.
The first attempt to compile our fork is actually merge mono SKIA to our repository, but that is too big to manage, and it also can't compile, not sure why. So, I try to build Google SKIA, and it compiled successfully. Based on that, I figure out how they do it. The changes on the first commit are just generated by applying a patch I created with their changes in gn/BUILD.gn file, the work should be credited to the Google team. The second commit is just putting back our changes to that file.
|
LGTM but I let owner of M1 approve and merge. |
|
Wow what a hero! Thank you @syaiful6 for bringing over this set of changes to unblock the M1 build 😍 |
|
Hoping to see this get merged soon -- This was something like step 5 (and the last step) on a chain of pull requests to get various projects working on apple silicon devices, the other each wait on the next up until this one =] |
|
Wow, hoping to see this get merged as well. Is there anything I can do to help this move along - test on M1 or something? |
hi all, this PR allow us to build SKIA in Mac M1 without merging upstream repository, this is minimal changes required. Gn tool set default cpu to x86, so we need to set it to arm64 if it in Darwin.
This PR will also make current Revery compiled natively in M1. For Onivim2 there is small changes needed, but it just a path of homebrew prefix in M1 which is /opt/homebrew.