Abstract LL(1) grammars have the conceptual and practical advantage that they allow the compiler writer to view the grammar as a program; this allows a more natural positioning of semantic actions and a simple attribute mechanism. Resulting parsers can be constructed that achieve fully automatic error‐recovery, which allows the compiler writer to ignore totally the issue of syntax errors. Measurement shows that such parsers can be reasonably efficient.

Abstract Requirements are formulated for a tag‐list algorithm, i.e. the algorithm used in a compiler for handling the symbol table or identifier list. Starting from a very general tag‐list algorithm, 18 practical versions are developed and their merits judged. Although the final choice (binary search in a diluted table) depends on the details of the application, the main part of this article is not devoted to that final choice itself but rather to ways of reaching it.