1 * MCS: The Ximian C# compiler
3 MCS began as an experiment to learn the features of C# by
4 writing a large C# program. MCS is currently able to parse C#
5 programs and create an internal tree representation of the
6 program. MCS can parse itself.
8 Work is progressing quickly on various fronts in the C#
9 compiler. Recently I started using the System.Reflection API
10 to load system type definitions and avoid self-population of
11 types in the compiler and dropped my internal Type
12 representation in favor of using the CLI's System.Type.
14 ** Phases of the compiler
16 The compiler has a number of phases:
19 * Lexical analyzer: hand-coded lexical analyzer that
20 provides tokens to the parser.
22 * The Parser: the parser is implemented using Jay (A
23 Berkeley Yacc port to Java, that I ported to C#).
24 The parser does minimal work and syntax checking,
25 and only constructs a parsed tree.
27 Each language element gets its own class. The code
28 convention is to use an uppercase name for the
29 language element. So a C# class and its associated
30 information is kept in a "Class" class, a "struct"
31 in a "Struct" class and so on. Statements derive
32 from the "Statement" class, and Expressions from the
35 * Parent class resolution: before the actual code
36 generation, we need to resolve the parents and
37 interfaces for interface, classe and struct
40 * Semantic analysis: since C# can not resolve in a
41 top-down pass what identifiers actually mean, we
42 have to postpone this decision until the above steps
45 * Code generation: nothing done so far, but I do not
46 expect this to be hard, as I will just use
47 System.Reflection.Emit to generate the code.
52 ** Current pending tasks
57 * Array declarations are currently being ignored,
59 * PInvoke declarations are not supported.
61 * Pre-processing is not supported.
63 * Attribute declarations and passing currently ignored.
65 * Compiler does not pass around line/col information from tokenizer for error reporting.
67 * Jay does not work correctly with `error'
68 productions, making parser errors hard to point. It
69 would be best to port the Bison-To-Java compiler to
70 become Bison-to-C# compiler (bjepson@oreilly.com
71 might have more information)
74 Interesting and Fun hacks to the compiler:
77 * Finishing the JB port from Java to C#. If you are
78 interested in working on this, please contact Brian
79 Jepson (bjepson at oreilly d-o-t com).
81 More on JB at: <a href="http://www.cs.colorado.edu/~dennis/software/jb.html">
82 http://www.cs.colorado.edu/~dennis/software/jb.html</a>
84 JB will allow us to move from the Berkeley Yacc
85 based Jay to a Bison-based compiler (better error
86 reporting and recovery).
88 * Semantic Analysis: Return path coverage and
89 initialization before use coverage are two great
90 features of C# that help reduce the number of bugs
91 in applications. It is one interesting hack.
93 * TypeRefManager. This exists currently in its infancy only.
95 * Enum resolutions: it is another fun hack, as enums can be defined
96 in terms of themselves (<tt>enum X { a = b + 1, b = 5 }</tt>).
100 ** Questions and Answers
102 Q: Why not write a C# front-end for GCC?
104 A: I wanted to learn about C#, and this was an exercise in this
105 task. The resulting compiler is highly object-oriented, which has
106 lead to a very nice, easy to follow and simple implementation of
109 I found that the design of this compiler is very similar to
110 Guavac's implementation.
112 Targeting the CIL/MSIL byte codes would require to re-architecting
113 GCC, as GCC is mostly designed to be used for register machines.
115 The GCC Java engine that generates Java byte codes cheats: it does
116 not use the GCC backend; it has a special backend just for Java, so
117 you can not really generate Java bytecodes from the other languages
120 Q: If your C# compiler is written in C#, how do you plan on getting
121 this working on a non-Microsoft environment.
123 We will do this through an implementation of the CLI Virtual
124 Execution System for Unix (our JIT engine).
128 A: No, currently I am using Jay which is a port of Berkeley Yacc to
129 Java that I later ported to C#. This means that error recovery is
130 not as nice as I would like to, and for some reason error
131 productions are not being caught.
133 In the future I want to port one of the Bison/Java ports to C# for
136 Q: Should someone work on a GCC front-end to C#?
138 A: I would love if someone does, and we would love to help anyone that
139 takes on that task, but we do not have the time or expertise to
140 build a C# compiler with the GCC engine. I find it a lot more fun
141 personally to work on C# on a C# compiler, which has an intrinsic
144 We can provide help and assistance to anyone who would like to work
147 Q: Should someone make a GCC backend that will generate CIL images?
149 A: I would love to see a backend to GCC that generates CIL images. It
150 would provide a ton of free compilers that would generate CIL
151 code. This is something that people would want to look into
152 anyways for Windows interoperation in the future.
154 Again, we would love to provide help and assistance to anyone
155 interested in working in such a project.
157 Q: What about making a front-end to GCC that takes CIL images and
158 generates native code?
160 A: I would love to see this, specially since GCC supports this same
161 feature for Java Byte Codes. You could use the metadata library
162 from Mono to read the byte codes (ie, this would be your
163 "front-end") and generate the trees that get passed to the
166 Ideally our implementation of the CLI will be available as a shared
167 library that could be linked with your application as its runtime
170 Again, we would love to provide help and assistance to anyone
171 interested in working in such a project.
173 Q: But would this work around the GPL in the GCC compiler and allow
174 people to work on non-free front-ends?
176 A: People can already do this by targeting the JVM byte codes (there
177 are about 130 compilers for various languages that target the JVM).
179 Q: Why are you writing a JIT engine instead of a front-end to GCC?
181 A: The JIT engine and runtime engine will be able to execute CIL
182 executables generated on Windows.
185 You might also want to look at the <a href="faq.html#gcc">GCC</a>
186 section on the main FAQ