Currently the quality (performance) of generated code does not degrade particularly gracefully as the quality of type-inference degrades. There are a number of different projects possible here. Some example optimization opportunities include:
Recommended Skills: Experience with reading LLVM IR or assembly. An understanding of formal logic and using it to develop improved code transformations.
Mentors: Jameson Nash
There are many remaining components that need to be updated to use thread-safe algorithms before Julia’s threading will be stable for general usage. Some basic data-structures (such as the TypeMap) are missing correct RCU and memory barriers to ensure race-free answers. IO is also currently unavailable for multi-threaded code. The realloc operation for arrays (i.e. resize!) may be more reliable if it was implemented using RCU malloc (delaying the free until a gc-safepoint has been reached on all threads).
Expected Results: Demonstrate that a program that fails at the beginning of the summer can now run to completion, or show what remains to be fixed to get it working. Start a test framework to help find bugs and race conditions in the current implementation, and to detect future regressions in multi-threading behavior.
Recommended Skills: Experience with thread-safety in C.
Mentors: Jameson Nash, Yichao Yu
Memory errors in Julia’s underlying C code are sometimes difficult to trace, and missing garbage-collector “roots” (GC roots) can lead to segfaults and other problems. One way to find such errors is to check Julia’s src/ directory for missing GC roots by determining which call chains can trigger garbage collection, and then looking for objects that lack GC root protection. Toward this end, a tool called ClangSA.jl was developed using Clang’s static analyzer (as well as the Cxx package) and has been used to find some GC bugs. (alternatively, coccinelle could be investigated)
Expected Results: Expand and improve the ClangSA tool; potentially provide continuous integration support (for example, a Nanosoldier command).
Required Skills: Familiarity with C, runtimes, and experience with (or interest in) static analysis.
Recommended Skills: Experience with Julia’s C runtime.
Mentors: Keno Fischer, Yichao Yu