afonso360 opened PR #4817 from interp-reg-v2 to main:
:wave: Hey,
Lets try this again! The issue in #4800 was due to the signature checking that we included in #4782 being overly strict.
In the interpreter our
DataValues do not have full type information about their data. We only have a single boolean representationBinstead of one for each sizeb1/b8/b16/etc.., this caused a error in the newcall.cliftests since when we query the types of theseDataValues we just return a "Default" type. Which in this case wasb8and different fromb1in the signature, therefore a signature error!The solution in the last commit is to relax our signature checking constraints for these two type categories.
cc: #4800
cc: #4810
afonso360 edited PR #4817 from interp-reg-v2 to main:
:wave: Hey,
Lets try this again! The issue in #4800 was due to the signature checking that we included in #4782 being overly strict.
In the interpreter our
DataValues do not have full type information about their data. We only have a single boolean representationBinstead of one for each sizeb1/b8/b16/etc.., this caused a error in the newcall.cliftests since when we query the types of theseDataValues we just return a "Default" type. Which in this case wasb8and different fromb1in the signature, therefore a signature error!The solution in the last commit is to relax our signature checking constraints for these two type categories (bools & vectors).
cc: #4800
cc: #4810
jameysharp submitted PR review.
jameysharp merged PR #4817.
Last updated: Jan 09 2026 at 13:15 UTC