Wrong error message in case of validation error

[vouillon]

Running this test program, I get the following error:

$ ./bin/wizeng.x86-linux  -ext:gc -ext:exception-handling ./print_array.wasm 
./print_array.wasm:0x3D04:Code[49]: expected 148 data segments, missing data section

We should get the following validation error instead:

 ./bin/wizeng.x86-linux  -ext:gc -ext:exception-handling ./print_array.wasm 
./print_array.wasm:0x7F82:Code[49]: expected type (ref array #0), got i32

The issue is that in case of validation error, the parsing is interrupted but the check for missing data section is still performed, overriding the previous error.

[titzer]

Ok, looks like this is due to the location where the error is being report. Generally, Wizard tries to report the error at the earliest byte. It could, e.g. report the missing data section at the end of the module, and any previous error would then have precedence. Wdyt?

[vouillon]

The data section is not missing. But since the parsing is interrupted before reaching the data section due to the validation error, the check for a missing data section fails. I think the check should be skipped if there is a pending error.