You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As note:
An alternative (to #165) way how this issue manifests is during code generation (instead of application):
%core.trait.* is expected to be resolved entirely before code generation.
In some cases, however, it remains and no size is computed.
Example: %core.trait.size «_3152207; «_3152209; .Idx 4294967296»».
Fork: Neuralcoder3
Branch: matrix_dialect
Call: ./build/bin/thorin -d matrix lit/matrix/print_const_dyn_mat.thorin -d affine -d direct -d clos -o - -VVVV --output-ll T.ll
Note: add_mem is disabled due to another bug (possibly related to projections)
The indices are statically unknown.
Note: Code generation fails with assert(false); without telling that it happens during code generation or what is wrong.
A more informative error like assert(false && "trait should be lowered before codegen"); would be helpful.
We've already mentioned that in a couple of other issues like #9 that we probably want to move away from assertions and instead use proper error messages/exceptions instead. In the midterm we could expose this "failure is an option" to the compilation pipeline.
What should we do with sth like this?
See #165.
The text was updated successfully, but these errors were encountered: