Conversation
|
Woo! Fortran! A small comment before my brain turns off completely - the change in 9e4ba1c about path separators appears as though it will be backwards-incompatible with |
|
I don't think it should be backwards-incompatible where it is currently works (Linux / macOS), and on Windows the old behavior was broken, so we don't need to worry about the change in behavior there, right? |
|
Yeah, I agree about macOS and Linux. Was it broken on Windows such that the user would get an error message and the build would stop? In that case, we don't need to worry I guess. But if it was able to work with colons instead of semicolons at all (in other words, no error message even if it ended up doing something incorrect), then I think it would be breaking in the sense that there is now a behavior change for the user (to the correct behavior, but still...). |
|
I added an error message to what I think is the only case where using a colon as a path separator in |
When using the MSVC toolchain together with Fortran compilers like Flang, MSVC should be used as the linker.
Some compilers (including Flang) output the message:
"Warning: ieee_inexact is signaling"
when running the Fortran sample programs. This output is a consequence of
F2003 standard compliance by some compilers, and can be avoided by letting
the program exit normally rather than using the 'stop' statement.
Codecov Report
@@ Coverage Diff @@
## master #874 +/- ##
==========================================
- Coverage 71.56% 70.54% -1.02%
==========================================
Files 372 375 +3
Lines 44510 45349 +839
==========================================
+ Hits 31853 31993 +140
- Misses 12657 13356 +699
Continue to review full report at Codecov.
|
bryanwweber
left a comment
There was a problem hiding this comment.
Thanks for adding the warning
Changes proposed in this pull request
Checklist
scons build&scons test) and unit tests address code coverage