Ondřej Čertík
Ondřej Čertík
I had to revert the fix, it failed exhaustive tests.
```fortran program test_trim_write implicit none character(len=20) :: tmp_line(3) integer :: iii, ii tmp_line = "Hello" ii = 3 write(*, '(g0)') (trim(tmp_line(iii)), iii=1, ii) end program test_trim_write ```
Let's make sure we have tests for all code reported in this issue, and all code that is not fixed yet reported in separate issues.
Do any of these reference tests change when we apply the fix from https://github.com/lfortran/lfortran/pull/8957?
Ok, in that case we can merge these any time, as they are in some sense independent of the actual fixes.
These changes are hard to review, I need to play with it by hand. It's really easy to make a mistake in the test runner. I think all we need...
Perfect, that looks good. Now add a test for this new error message into continue_compilation_1.f90 and update reference tests.
Currently this is using try/except and I don't like that. I would prefer a solution that doesn't need to try/except, but let's at least check if CI passes with this....
Looks like everything passes. But I don't want to merge it, as I don't like the try/catch. Rather, we need to save the variable to a local temporary until we...
So far I haven't figured out a better solution without a large refactor -- the two use cases to covert: self-referential initialization (the variable must be in a symbol table...