Fix versions for Libs subset's native artifacts #52799
Conversation
During the repo consolidation, we missed passing `OfficialBuildId` argument to common script from all subsets, which resulted in default version emitted in the binaries produced by Libs subset in the official builds.
|
I couldn't figure out the best area label to add to this PR. If you have write-permissions please help me learn by adding exactly one area label. |
|
cc @janvorli before
... after
... |
|
CI failures are unrelated to PR changes. |
|
@am11 do we have the same issue in 5.0 too? |
|
@janvorli, yes, 5.0 has this problem as well and it was a regression from 3.1 where we have correct version in all ELF files produced by official build; shared object and executable. ebdcb5e fixed it in executable files, this PR fixes the remaining shared objects. |
|
@am11 thank you for the details, I'll create a porting PR after this is merged in. |
|
/backport to release/5.0 |
|
Started backporting to release/5.0: /~https://github.com/dotnet/runtime/actions/runs/853353504 |
|
I think this causes an issue on the runtime-staging pipeline in I've asked @radekdoulik to help investigate. |
|
should be fixed by #52917 |
|
Started backporting to release/5.0: /~https://github.com/dotnet/runtime/actions/runs/853353504 |
|
Hmm for some reason it thinks you're not a contributor: |
|
Fixed the backport bot issue with #52938 but I think we'll need to do a manual backport anyway since there were a few follow-ups. |
Port of dotnet#52799 and dotnet#52917 During the repo consolidation, we missed passing OfficialBuildId argument to common script from all subsets, which resulted in default version emitted in the binaries produced by Libs subset in the official builds.
|
@am11 I've created a port to release/5.0. I've actually found that only the libraries change was needed, the coreclr and installer binaries including the singlefilehost had a correct version. I can see that there was some change in the installer repo in 6.0 related to the versions, so that probably broke the installer. |
|
@janvorli, thank you. I think the singlefilehost patch would be good to add (the file was moved by the same change applies). Currently, on Ubuntu commit info is there but version is the placeholder (42.42.42..) in 5.0 installed with |
|
@am11 - that is strange, the version is correct in my local build that I've just produced using |
|
I ran Perhaps it is fixed for 5.0.7, I will double check once it is released. |
|
@ViktorHofer, it gets ultimately passed to empty.csproj here: runtime/eng/native/build-commons.sh Line 116 in 0dd6b7b which generates version.c file under obj directory (via arcade versioning targets). Currently, we emit sccid only in Unix binaries, and RC resource on windows based on the info derived from OfficialBuildId and version control ID (commit hash). Vitek's idea was to also emit sccid in Windows binary (as it simply improves diadnostibility in headless environments), which would requires some changes in the arcade target. |
* [Release/5.0] - Fix versions for Libs native artifacts Port of #52799 and #52917 During the repo consolidation, we missed passing OfficialBuildId argument to common script from all subsets, which resulted in default version emitted in the binaries produced by Libs subset in the official builds. * Fix empty build version case
ghost
locked as resolved and limited conversation to collaborators
During the repo consolidation, we missed passing
OfficialBuildIdargument to common script from all subsets, which resulted in default
version emitted in the binaries produced by Libs subset in the official
builds.