Cirrus has native ARM64 Linux jobs, so we should use them (see also #707 (comment))
And if we have that, and since MacOS jobs will run on ARM / M1 after #1152, maybe it makes sense to reduce the number of MacOS jobs a little bit. They really need a long time to start.
This "half-depends" on the support of Docker as a CI environment on ARM, see https://github.com/cirruslabs/cirrus-ci-docs/pull/905#issuecomment-1330488930
If the Cirrus ARM is a full vm, you could also install docker and then do the extra step of loading the dockerfile in the cirrus yml?
If the Cirrus ARM is a full vm, you could also install docker and then do the extra step of loading the dockerfile in the cirrus yml?
In principle yes, though the main advantage of the cirrus magic is to manage caching and rebuilding of the docker image. Caching speeds up our builds a lot because we have so many different builds.
This "half-depends" on the support of Docker as a CI environment on ARM, see cirruslabs/cirrus-ci-docs#905 (comment)
This should work now according to https://github.com/cirruslabs/cirrus-ci-docs/issues/1154.