-
-
Notifications
You must be signed in to change notification settings - Fork 5.6k
Comparing changes
Open a pull request
base repository: JuliaLang/julia
base: master@{1day}
head repository: JuliaLang/julia
compare: master
- 8 commits
- 25 files changed
- 4 contributors
Commits on Sep 12, 2025
-
Configuration menu - View commit details
-
Copy full SHA for 7510b54 - Browse repository at this point
Copy the full SHA 7510b54View commit details -
Configuration menu - View commit details
-
Copy full SHA for 1f463c3 - Browse repository at this point
Copy the full SHA 1f463c3View commit details -
Configuration menu - View commit details
-
Copy full SHA for b306585 - Browse repository at this point
Copy the full SHA b306585View commit details
Commits on Sep 16, 2025
-
optimizations: better modeling and codegen for apply calls
- Use svec instead of tuple for arguments (better match for ABI which will require boxes) - Directly forward single svec argument, both runtime and codegen, without copying. - Optimize all consistant builtin functions of constant arguments, not just ones with special tfuncs. Reducing code duplication and divergence. - Codegen for `svec()` directly, so optimizer can see each store (and doesn't have to build the whole thing on the stack first).
Configuration menu - View commit details
-
Copy full SHA for 19ad3be - Browse repository at this point
Copy the full SHA 19ad3beView commit details -
codegen: mark svec fields GC as release stores
Without a release store, it seems LLVM considers it a data race to have read the initial state on another thread. Marking this as a release store seems sufficient to prevent that optimization. It is also more consistent with how we initialize and write to most other structs, particularly since #55767. Fixes #59547 (more)
Configuration menu - View commit details
-
Copy full SHA for 18f1c26 - Browse repository at this point
Copy the full SHA 18f1c26View commit details
Commits on Sep 17, 2025
-
optimizations: better modeling and codegen for apply and svec calls (#…
…59548) - Use svec instead of tuple for arguments (better match for ABI which will require boxes) - Directly forward single svec argument, both runtime and codegen, without copying. - Optimize all consistant builtin functions of constant arguments, not just ones with special tfuncs. Reducing code duplication and divergence. - Codegen for `svec()` directly, so optimizer can see each store (and doesn't have to build the whole thing on the stack first). Written with help by Claude
Configuration menu - View commit details
-
Copy full SHA for f818842 - Browse repository at this point
Copy the full SHA f818842View commit details -
move rem2pi and pow to special (#59545)
This makes math.jl slightly simpler with no functional differences. While cleaning up, I also removed Base.Math.libm which appears to be completely unused (and is an alias to Base.libm_name)
Configuration menu - View commit details
-
Copy full SHA for e0d49ac - Browse repository at this point
Copy the full SHA e0d49acView commit details -
prevent stack overflows from faulty user-defined constructor methods (#…
…59506) `Base` code often assumes that a constructor will return a value of the stated type. This is false, cross-reference issue #42372. This change prevents several stack overflows that arise when a generic method is called with a newly-defined type with such a weird constructor for some `Base`-defined type.
1Configuration menu - View commit details
-
Copy full SHA for f0ece4a - Browse repository at this point
Copy the full SHA f0ece4aView commit details
This comparison is taking too long to generate.
Unfortunately it looks like we can’t render this comparison for you right now. It might be too big, or there might be something weird with your repository.
You can try running this command locally to see the comparison on your machine:
git diff master@{1day}...master