-
Notifications
You must be signed in to change notification settings - Fork 754
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
[SYCL][CUDA] Make plugin specific error return an error #10626
[SYCL][CUDA] Make plugin specific error return an error #10626
Conversation
Most of your description should be in the commit message proper. Code looks good to me. Also, because of how this project has been configured, your mention of steffen will end up in project history. I've tried arguing against that without success, but for now you should treat descriptions on github as if they will end up in git history (because they will) |
(Editing after @ldrumm has mentioned we should move conversation out of PR description to avoid contaminating the commit message.) I am not sure if this is desirable behaviour. @steffenlarsen what do you think? I have also added a test to check that the |
What behavior? @ldrumm is right that the description is what normally ends up as the commit message as that's what is used when we squash the commits. This is the intended behavior, so it would indeed be preferable if you keep the description in line with how you want the final commit to appear and avoid tagging inside it. |
Sorry I meant the behaviour introduced by this PR. I have actually changed the behaviour so As a sidenote I think for clarity it might be nice to split |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good catch! I believe this was indeed the intended behavior. Tag @smaslov-intel on his thoughts as well as for the split error code for distinguishing between warnings and errors.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
👍 for error/warning split
@intel/llvm-gatekeepers This can be merged. I'll make a new ticket to track the |
@intel/dpcpp-l0-pi-reviewers approval is missing. Friendly ping. |
@intel/llvm-gatekeepers can this now be merged? |
The `UR_RESULT_ADAPTER_SPECIFIC_ERROR` was not returning an error to the SYCL RT which meant all errors were treated as warnings and ignored unless `SYCL_RT_WARNING_LEVEL` is set to geq 2. This changes things so the adapter specific error is now reported as such, meaning all uses `UR_RESULT_ADAPTER_SPECIFIC_ERROR` meant as warnings are now caught as errors. --------- Co-authored-by: Hugh Delaney <[email protected]>
The `UR_RESULT_ADAPTER_SPECIFIC_ERROR` was not returning an error to the SYCL RT which meant all errors were treated as warnings and ignored unless `SYCL_RT_WARNING_LEVEL` is set to geq 2. This changes things so the adapter specific error is now reported as such, meaning all uses `UR_RESULT_ADAPTER_SPECIFIC_ERROR` meant as warnings are now caught as errors. --------- Co-authored-by: Hugh Delaney <[email protected]>
The
UR_RESULT_ADAPTER_SPECIFIC_ERROR
was not returning an error to the SYCL RT which meant all errors were treated as warnings and ignored unlessSYCL_RT_WARNING_LEVEL
is set to geq 2. This changes things so the adapter specific error is now reported as such, meaning all usesUR_RESULT_ADAPTER_SPECIFIC_ERROR
meant as warnings are now caught as errors.