Added partial support for no-std environments

[KernelFreeze]

Description

Added dynamic imports and feature flags to allow support for no-std environments in wasmer-compiler and singlepass-compiler as a first step for supporting no-std in Wasmer.

Review

• Add a short description of the change to the CHANGELOG.md file

[Hywan]

Thanks for your patch, I'll review it soon :-).

[MarkMcCaskey]

I skimmed through the changes and they seem reasonable -- thanks for the PR!

For some background, we haven't been working toward keeping Wasmer no-std supporting because we don't have a clear use case for it yet and it adds some overhead to the code we maintain. That said, if there are use cases for no-std Wasmer then we'd like to support it. And at least so far, the changes in this PR don't seem too burdensome.

On a related note to that, I think we'll need a test of some kind in CI or we'll likely break no-std support the day after this PR lands. A test could be a good opportunity to show-case some of the uses of no-std support, though that may be harder than it initially seems. I suspect we may run into issues with some of the OS-specific stuff in wasmer-vm, for example, when trying to use Wasmer with no-std. To not block this PR on something as (potentially) hard as getting no-std wasmer working in a test, perhaps we can just test that these crates compile with the no-std feature enabled (cargo check) or that the unit tests of these crates pass with no-std.

I'd like to hear your thoughts on this!

[KernelFreeze]

For some background, we haven't been working toward keeping Wasmer no-std supporting because we don't have a clear use case for it yet and it adds some overhead to the code we maintain. That said, if there are use cases for no-std Wasmer then we'd like to support it. And at least so far, the changes in this PR don't seem too burdensome.

Hey, thank you for your quick reply! I have a use case that can be quite interesting, running Wasmer in a Rust Kernel, creating an environment to run WASM in bare-metal hardware. See etheryal-kernel for more details.

On a related note to that, I think we'll need a test of some kind in CI or we'll likely break no-std support the day after this PR lands. A test could be a good opportunity to show-case some of the uses of no-std support, though that may be harder than it initially seems. I suspect we may run into issues with some of the OS-specific stuff in wasmer-vm, for example, when trying to use Wasmer with no-std.

I can try to add an integration test of some sort using no-std, if not possible, we could just make it compile with core feature flag, so it disables any std integration.

[syrusakbary]

I can try to add an integration test of some sort using no-std, if not possible, we could just make it compile with core feature flag, so it disables any std integration.

Sounds good!

[syrusakbary]

@KernelFreeze the progress is looking great. Let us know once the PR is ready to review :)

[KernelFreeze]

Sure! Thank you :)

[syrusakbary]

Hey @KernelFreeze, we are reviewing stalled PRs and it would be great to know if you want to continue working on this

[KernelFreeze]

Hey! I don't have time at the moment, so I will reopen another pull request when I'm ready. Thank you!