\chapter{\label{modifying-chapter}Modifying the Compiler}
The compiler consists of about 30,000 lines of yacc, lex and C code
(another 7,000+ for the runtime library routines). The best way to
understand the compiler internals is to understand the module data
structure ({\ufn \dots/compiler/core/asn1module.h}) and to read the compiler
chapter in this document to gain a conceptual understanding of each
pass of the compiler.
The most common form of modification will likely be for macro
handling. To understand this, look at the way the OBJECT-TYPE macro is
treated in:
\begin{description}
\item[lex-asn1.l] {add any new keywords}
\item[parse-asn1.y] { parse the macro into the desired data structure.
Use the existing productions as much as possible.}
\item[link-type.c] { link any type defined or referenced in the
macro}
\item[link-values.c] { link any value defined or referenced in the
macro}
\item[do-macros.c] { perform any semantic action for the macro }
\item[normalize.c] { move any type and value definitions in the macro
to the top level so the code generator can generate code for them
(without looking in the macro).}
\item[code generators] { to convert any special semantics into useful
C or C++. This phase is likely to be dependent on the generated
code's target environment.}
\end{description}
In general I have tried to put comments where funky things happen and
to use function and variable names that are meaningful. However,
things may get ugly in certain places. Thesis writing is harmful to
your coding style!