teavm/tests
Alexey Andreev a2a9dbcbe3 Improve performance of inliner in FULL optimization level
The former implementation suffered from class inference.
The reason was in many nodes having too many possible classes in them.
The new implementation does not handle full set of classes in each node.
Instead, it introduces concept of 'overflow', i.e. node having
more types than the given upper limit.
These nodes behave as if there were all possible classes in them,
which allows to apply certain optimization for these nodes and
omit heavy computations of large type sets.
2019-09-27 16:36:55 +03:00
..
src/test Improve performance of inliner in FULL optimization level 2019-09-27 16:36:55 +03:00
compile-c-unix-fast.sh C: add option to support exceptions via setjmp/longjmp 2019-07-22 16:53:47 +03:00
compile-c-unix.sh C: compile into multiple .c and .h files 2019-05-08 17:36:46 +03:00
pom.xml Update version to 0.7.0 2019-09-24 16:26:14 +03:00