Improve type inference of non-const GlobalRef's
#168
Conversation
|
Needs a test, but otherwise looks good. |
staticfloat
force-pushed
the
sf/globalref_types
branch
from
957d06d to
72dc8e7
Compare
|
Test added, luckily it was not as difficult as I feared. I have verified that the test fails without this change. Unfortunately, the reverse mode tests were broken by JuliaLang/julia@ec33194 and that might be a heavy lift to fix. We'll check in with Keno tomorrow to see how bad it's going to be to fix that (he says it's a miscompile, which sounds bad). If it's too too painful we can either disable those reverse-mode tests (since we're not using them anyway), look into some kind of workaround, or even merge this with the reverse mode tests failing. |
GlobalRefs often do not get their types refined, so we should be careful to re-use types that we have when possible. This PR moves non-`const` GlobalRef handling out of `maparg()` so that we can keep the original type.
oxinabox
force-pushed
the
sf/globalref_types
branch
from
72dc8e7 to
e0963a9
Compare
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## main #168 +/- ##
=======================================
Coverage 22.96% 22.96%
=======================================
Files 27 27
Lines 2661 2665 +4
=======================================
+ Hits 611 612 +1
- Misses 2050 2053 +3
☔ View full report in Codecov by Sentry. |
GlobalRefs often do not get their types refined, so we should be careful to re-use types that we have when possible. This PR moves non-
constGlobalRef handling out ofmaparg()so that we can keep the original type.