Abstract. We describe support for modularity in Newspeak, a programming language descended from Smalltalk [33] and Self [69]. Like Self, all computation -even an object's own access to its internal structureis performed by invoking methods on objects. However, like Smalltalk, Newspeak is class-based. Classes can be nested arbitrarily, as in Beta [44]. Since all names denote method invocations, all classes are virtual; in particular, superclasses are virtual, so all classes act as mixins. Unlike its predecessors, there is no static state in Newspeak, nor is there a global namespace. Modularity in Newspeak is based exclusively on class nesting. There are no separate modularity constructs such as packages. Top level classes act as module definitions, which are independent, immutable, self-contained parametric namespaces. They can be instantiated into modules which may be stateful and mutually recursive.

Live programming, originally introduced by Smalltalk and Lisp, and now gaining popularity in contemporary systems such as Swift, requires on-the-fly support for object schema migration, such that the layout of objects may be changed while the program is at one and the same time being run and developed. In Smalltalk schema migration is supported by two primitives, one that answers a collection of all instances of a class, and one that exchanges the identities of pairs of objects, called the become primitive. Existing instances are collected, copies using the new schema created, state copied from old to new, and the two exchanged with become, effecting the schema migration.Historically the implementation of become has either required an extra level of indirection between an object's address and its body, slowing down slot access, or has required a sweep of all objects, a very slow operation on large heaps. Spur, a new object representation and memory manager for Smalltalk-like languages, has neither of these deficiencies. It uses direct pointers but still provides a fast become operation in large heaps, thanks to forwarding objects that when read conceptually answer another object and a partial read barrier that avoids the cost of explicitly checking for forwarding objects on the vast majority of object accesses.

OpenSmalltalk-VM is a virtual machine (VM) for languages in the Smalltalk family (e.g. Squeak, Pharo) which is itself written in a subset of Smalltalk that can easily be translated to C. Development is done in Smalltalk, an activity we call "Simulation". The production VM is derived by translating the core VM code to C. As a result, two execution models coexist: simulation, where the Smalltalk code is executed on top of a Smalltalk VM, and production, where the same code is compiled to an executable through a C compiler. In this paper, we detail the VM simulation infrastructure and we report our experience developing and debugging the garbage collector and the just-in-time compiler (JIT) within it. Then, we discuss how we use the simulation infrastructure to perform analysis on the runtime, directing some design decisions we have made to tune VM performance. CCS Concepts • Software and its engineering → Runtime environments; Just-in-time compilers; Interpreters;

Speculative inlining in just-in-time compilers enables many performance optimizations. However, it also introduces significant complexity. The compiler optimizations themselves, as well as the deoptimization mechanism are complex and error prone. To stabilize our bytecode to bytecode just-in-time compiler, we designed a new approach to validate the correctness of dynamic deoptimization. The approach consists of the symbolic execution of an optimized and an unoptimized bytecode compiled method side by side, deoptimizing the abstract stack at each deoptimization point (where dynamic deoptimization is possible) and comparing the deoptimized and unoptimized abstract stack to detect bugs. The implementation of our approach generated tests for several hundred thousands of methods, which are now available to be run automatically after each commit.

No abstract