-
Notifications
You must be signed in to change notification settings - Fork 124
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
v15 regression: temporary confusion with an inout (validating too hard) #1455
Comments
Plot twists, if all validation is deactivated the code compiles, runs and validates. Seems like overvalidation. Changing the issue title. |
Looks like an input (tmp35) is not added to the nested SDFG as an input, but is used as one. So validation fails. @alexnick83 any ideas on why something would be output-only? |
I had a quick look at the SDFG and I cannot find the issue reported by the validation routine. The error concerns inout connector @tbennun |
That makes sense. I propose we change validation, not behavior As for validation, there are two methods ( |
This is the commit that added them: 4b14a73 |
I see the problem. I haven't considered that case. I will make a PR. |
PR is up. I tested the SDFG and it passes validation. |
Previously working code doesn't pass validation anymore. It seems the code gets confused between a temporary it created and the original temporary field
Minimal reproducer is to run the Remapping numerical regression test on either cpu or gpu backend. Below a script to run on CPU (DaCe version is post #1446)
Attached is the invalid.sdfgz (renamed in .zip because github)
The text was updated successfully, but these errors were encountered: