Skip to content

[release/10.0.2xx] ProcessFrameworkReferences: use portable ILCompiler when RuntimeIdentifier is null.#52486

Merged
baronfel merged 7 commits intorelease/10.0.2xxfrom
backport/pr-52407-to-release/10.0.2xx
Jan 27, 2026
Merged

[release/10.0.2xx] ProcessFrameworkReferences: use portable ILCompiler when RuntimeIdentifier is null.#52486
baronfel merged 7 commits intorelease/10.0.2xxfrom
backport/pr-52407-to-release/10.0.2xx

Conversation

@github-actions
Copy link
Contributor

@github-actions github-actions bot commented Jan 15, 2026

Backport of #52407 to release/10.0.2xx

Fixes #52395
Fixes #42335

Customer Issue

Users trying to publish NativeAOT applications on source-built SDKs would error because the tooling always directed them to non-portable versions of the NativeAOT tooling - which may not be present in the installs. The desired semantic is that the source-built AOT tooling should only be used when targeting the non-portable RID associated with the source-built SDK platform.

Description

This PR ensures that on all SDKs, the 'portable' version of the AOT tooling is used when no explicit RID is chosen for targeting. This means that by default applications that users build with NativeAOT will user Microsoft binaries from NuGet. Users that specifically want to target non-portable RIDs (e.g. fedora.x.y) must specify that RID explicitly.

Was this a regression?

  • Yes - it was introduced in a codeflow PR accidentally.
  • No

Testing

A new test has been added that explicitly maps out the expected behaviors:

  • no RID specified -> portable tooling selected
  • portable RID specified -> portable tooling selected
  • non-portable RID specified -> non-portable tooling selected

Risk

Low - there is broader testing coverage of asset selection now and this pathway continues to work as expected for all other AOT/Trimming related publishing tests.

@mmitche mmitche deleted the branch release/10.0.2xx January 17, 2026 00:38
@mmitche mmitche closed this Jan 17, 2026
@baronfel baronfel reopened this Jan 20, 2026
@baronfel baronfel self-assigned this Jan 20, 2026
…times if it's empty (#52685)

Co-authored-by: Marc Paine <marcpop@microsoft.com>
@baronfel
Copy link
Member

/ba-g known template engine macos ssl error

@baronfel baronfel merged commit 918016b into release/10.0.2xx Jan 27, 2026
22 of 29 checks passed
@baronfel baronfel deleted the backport/pr-52407-to-release/10.0.2xx branch January 27, 2026 20:02
@dsplaisted
Copy link
Member

@baronfel FYI, it looks like in addition to the templating issue on this one there were some leg timeouts which we've started seeing, not sure what's going on with them yet.

@baronfel
Copy link
Member

Oh good grief, and because they were canceled they didn't block the merge?

1 similar comment
@baronfel
Copy link
Member

Oh good grief, and because they were canceled they didn't block the merge?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants

Comments