This paper describes a new LLVM-based backend for the Parallel ML compiler (part of the Manticore system). This backend is novel in that it supports heap-allocated first-class continuations (a first for LLVM), which, in turn enables language features, such as
callcc, lightweight concurrency mechanisms, and PML’s parallelism features.
Thu 22 Sep
|11:45 - 12:10|
|12:10 - 12:25|
Tomohiro SasakiTohoku University, Katsuhiro UenoTohoku University, Japan, Atsushi OhoriTohoku University, JapanPre-print