Stream: git-wasmtime

Topic: wasmtime / Issue #1681 Better printing of V128 values in ...

Wasmtime GitHub notifications bot (May 11 2020 at 20:45):

abrown opened Issue #1681:

When troubleshooting WAST test failures, it is difficult to determine what is different between the expected V128 pattern (e.g. [0, 1, 2...]) and the giant integer contained in Val::V128. I attempted a solution to this in #1650 that changed the internal Rust representation of V128 values from u128 to [u8; 16]. Though this worked and all tests passed, it was only a partial solution: with [u8; 16] it was visually easier to troubleshoot which lanes were incorrect but it didn't exactly match the output of, e.g., an i32x4 vector.

To fully resolve this, we need to either manually implement Debug for Val or add a ValDisplay struct to:

• print V128 values as hexadecimal (and do the same with the expected side in wast.rs)
• or use the variants of V128Pattern to parse and display the Val::V128 to match the V128Pattern lanes/types.

Wasmtime GitHub notifications bot (Dec 04 2020 at 22:44):

abrown closed Issue #1681:

When troubleshooting WAST test failures, it is difficult to determine what is different between the expected V128 pattern (e.g. [0, 1, 2...]) and the giant integer contained in Val::V128. I attempted a solution to this in #1650 that changed the internal Rust representation of V128 values from u128 to [u8; 16]. Though this worked and all tests passed, it was only a partial solution: with [u8; 16] it was visually easier to troubleshoot which lanes were incorrect but it didn't exactly match the output of, e.g., an i32x4 vector.

To fully resolve this, we need to either manually implement Debug for Val or add a ValDisplay struct to:

• print V128 values as hexadecimal (and do the same with the expected side in wast.rs)
• or use the variants of V128Pattern to parse and display the Val::V128 to match the V128Pattern lanes/types.

Last updated: Nov 22 2024 at 16:03 UTC