One should consider creating a VM first - it simplifies life a bit, and makes the language more portable in the future. The implementation task is split in two

One should consider creating a VM first - it simplifies life a bit, and makes the language more portable in the future. The implementation task is split in two - compiler and VM. First, compiler targets the VM. Second, the VM targets the CPU. This way porting to different architectures is just a matter of adding to the VM to interface with the new arch. A compiler and VM definitely are 2 separate projects with their own set of issues. Also we should consider the possibility of going the eDSL (embedded domain-specific language) route. Working at a very high level, re-using existing constructs of the host language. If one is interested, looking into SICP [1] will show an example of such a project. Using Scheme to build an interpreter. This can easily be done in several days. Best to try to reduce doing work that has been done many times before. ---------- [1] https://mitpress.mit.edu/sites/default/files/sicp/index.html https://mitpress.mit.edu/sites/default/files/sicp/index.html