A small, but important, concurrency problem in Verilog's semantics? (Work in progress)

Despite its many flaws, Verilog is today both the most popular hardware design language and a popular language for communication between hardware development tools. Ever since the language was standardised, researchers have made attempts at formalising its semantics. To this day, no such attempt has been fully successful. In this paper, we highlight one - we think, important - concurrency problem in Verilog's semantics that has, for now, sidetracked our own ongoing Verilog semantics formalisation attempt. To us, the problem calls for a clarification of the Verilog standard. We propose a potential fix for the problem.