Drop "ifdef GHC"s on INLINABLE pragmas #1167
Conversation
|
Motivated by #1163 (comment) |
|
Aren't they there because |
|
That's a good point to consider, here's the report on pragmas: https://www.haskell.org/onlinereport/haskell2010/haskellch12.html
So, if the non-GHC compiler follows the report there should be no problems with unguarded INLINABLE pragmas. |
|
They'll very likely throw up a slew of warnings 🫤. That hypothetical problem is probably worth incurring to remove the actual clutter though. |
|
Yes, I also think cleaning up the clutter would be nice. For what it's worth, MHS is the only such compiler at the moment and it's not complaining. |
These were added long ago in 06e91ac as an improvement on GHC over the already existing INLINE pragmas. We can get rid of these unnecessary (and confusing to contributors) ifdefs today since we do not have other compilers where we want the non-INLINABLE behavior to apply. While INLINABLE is not a standard pragma listed in the Haskell 2010 report, the report also specifies that unrecognized pragmas should be ignored. So we don't expect this to cause issues for non-GHC compilers. MicroHs quietly ignores them as expected.
meooow25
force-pushed
the
remove-ifdefghcs
branch
from
9740948 to
7c682ae
Compare
2 participants
These were added long ago in 06e91ac as an improvement on the already existing INLINE pragmas. We can get rid of these unnecessary (and confusing to contributors) ifdefs today since we do not have other compilers where we want the non-INLINABLE behavior to apply.