Details
Diff Detail
- Repository
- rL LLVM
Event Timeline
ELF/Relocations.cpp | ||
---|---|---|
865–866 ↗ | (On Diff #118305) | Can you just continue without checking for ErrorCount? |
ELF/Relocations.cpp | ||
---|---|---|
865–866 ↗ | (On Diff #118305) | Not quite :( Testing Time: 2.76s Failing Tests (36): |
ELF/Relocations.cpp | ||
---|---|---|
865–866 ↗ | (On Diff #118305) | Why are they failing? |
ELF/Relocations.cpp | ||
---|---|---|
865–866 ↗ | (On Diff #118305) | Because recordUndefined is not a function returning a boolean which communicates to the caller whether the reported symbol was actually undefined, it just does side effect calling error (or warn). |
test/ELF/pr34872.s | ||
---|---|---|
8 ↗ | (On Diff #118305) | Testing full error message with NOT is probably brittle and |
test/ELF/pr34872.s | ||
---|---|---|
8 ↗ | (On Diff #118305) | it's unlikely this message will change imho. |
test/ELF/pr34872.s | ||
---|---|---|
8 ↗ | (On Diff #118305) | +1 for @grimar's suggestion. We cannot predict how the error messages might change in the future: even if it's unlikely, it's still possible that they might for some reason (e.g. we might decide to do a wholesale change in their structure, like the move to change to multi-line errors for duplicate symbols). |
Fine, I'll change the test. Any other comments or this can go in once the test is fixed? @ruiu ?