-
Notifications
You must be signed in to change notification settings - Fork 200
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Developers can use native AOT experiment to target Windows and Linux ARM64 #336
Comments
@agocke can you parent this under the right theme |
@jeffschwMSFT Since this is a User Story, I added it to the #248 Epic. Does that look right to you? |
ARM64 support is great but there is a lack of a user story here to match the ambitions of the title ("widen platform compatibility"). Here are some major .Net platforms with the current state of support:
|
We use the term "platform" to refer to OS+arch, these are what I would call "app frameworks" At the moment we're not working on broadening app framework support, but focusing on traditional command line-like apps, while broadening support for more underlying platforms. |
@agocke we are trying to fix all user story titles to be in terms of the user visible result. Is this a reasonable title: "Developers can use AOT to target Windows and Linux ARM64" ? |
Developers can use native AOT experiment to target Windows and Linux ARM64 |
It's valid to substitute "irrelevant" for "traditional command line-like" in the above sentence and it shows that he strategy is insane. There was a native aot framework (UWP), which was deployed in user-facing apps, has regressed with dotnet5, and you are saying that even after significant future development native AOT will still be unusable in practice. |
So happy with the direction this is all going in. Just a quick question from a customer who's been very happy with CoreRT on Intel Macs so far (and so are all of our customers in turn). Is there a particular reason why macOS was left out of this story? |
The scope of this user story is tailored to the needs of customers we are working closely with on this project. Once we have Windows and Linux ARM64 and macOS ARM64 works in mainline CoreCLR, I should be fairly trivial to enable macOS ARM64 as well. We will be happy to accept PR with the required changes and provide guidance along the way. |
Thank you, Jan! Really appreciate it. |
Does this mean the whole AOT thing here is just working for console applications and not for WPF application? If the answer is simply "yes" then I could save a lot of time wasting to read all guides, digging myself through the build process and fail on testing AOT on my WPF application. Also if the answer is "yes" you might leave a note here: https://github.com/dotnet/runtimelab/blob/feature/NativeAOT/docs/workflow/building/coreclr/nativeaot.md |
Is this still the idea? UWP is unusable without AOT. So this basically means developers who want performant UWP apps will need to use other frameworks than .NET? |
Would like to have it noted whether the resulting output binaries are also targeted for Alpine, or depend on glibc somehow. I assume that Alpine would be supported. |
We just enabled Alpine builds. Please give it a try and file issues on any problems you run into. The resulting binaries depend on either glibc or musl (same as regular .NET runtime), so you have to have a different native binary for Alpine ie |
This work appears to be largely complete, modulo any bugs we find |
Right now Native AOT runs on significantly fewer platforms than CoreCLR. We think that extending support to ARM64, which is already supported by CoreCLR, would be useful in identifying costs and benefits of the Native AOT form factor.
We'd like to add compatibility for Windows and Linux ARM64 to the Native AOT experiment and bring the experience closer to traditional CoreCLR form factors.
Work tracking
runtimelab Infrastructure
runtimelab Windows, Linux x64 bring-up
Archive CoreRT Repo
Improve reflection experience
Debugging
DumpLog
,DumpHeap
,VerifyHeap
andDumpAsync
commandsARM64 Compatibility
Improved Linux tracing
The text was updated successfully, but these errors were encountered: