1 * MCS: The Ximian C# compiler
3 MCS is currently able to compile itself and many more C#
4 programs (there is a test suite included that you can use).
5 It is routinely used to compile Mono, roughly half a million
8 We are in feature completion mode right now. There are still
9 a couple of areas that are not covered by the Mono compiler, but
10 they are very very few at this point (security attributes),
11 you can also browse the MCS <a href="http://bugzilla.ximian.com/buglist.cgi?product=Mono%2FMCS&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&email1=&emailtype1=substring&emailassigned_to1=1&email2=&emailtype2=substring&emailreporter2=1&changedin=&chfieldfrom=&chfieldto=Now&chfieldvalue=&short_desc=&short_desc_type=substring&long_desc=&long_desc_type=substring&bug_file_loc=&bug_file_loc_type=substring&keywords=&keywords_type=anywords&op_sys_details=&op_sys_details_type=substring&version_details=&version_details_type=substring&cmdtype=doit&newqueryname=&order=Reuse+same+sort+as+last+time&form_name=query">bugs</a> from Bugzilla.
13 A test suite is maintained to track the progress of
14 the compiler and various programs are routinely compiled and
19 Slides for the Mono C# Compiler presentation at .NET ONE are
21 href="http://primates.ximian.com/~miguel/slides-europe-nov-2002/Mono_C_Sharp_Overview_1007.sxi">here</a>
26 The Mono C# compiler is part of the `mcs' module in the Mono CVS
27 you can get it from our <a href="anoncvs.html">Anonymous CVS</a> server,
28 or you can get nightly <a href="download.html">download page</a>.
32 MCS is written in C# and uses heavily the .NET APIs. MCS runs
33 on Linux with the Mono runtime and on Windows with both the
34 .NET runtime and the Mono runtime.
36 ** Reporting Bugs in MCS
38 When you report a bug, try to provide a small test case that would
39 show the error so we can include this as part of the Mono C# regression
42 If the bug is an error or a warning that we do not flag, write
43 a sample program called `csXXXX.cs' where XXXX is the code number
44 that is used by the Microsoft C# compiler that illustrates the
45 problem. That way we can also do regression tests on the invalid
48 ** Phases of the compiler
50 The compiler has a number of phases:
53 * Lexical analyzer: hand-coded lexical analyzer that
54 provides tokens to the parser.
56 * The Parser: the parser is implemented using Jay (A
57 Berkeley Yacc port to Java, that I ported to C#).
58 The parser does minimal work and syntax checking,
59 and only constructs a parsed tree.
61 Each language element gets its own class. The code
62 convention is to use an uppercase name for the
63 language element. So a C# class and its associated
64 information is kept in a "Class" class, a "struct"
65 in a "Struct" class and so on. Statements derive
66 from the "Statement" class, and Expressions from the
69 * Parent class resolution: before the actual code
70 generation, we need to resolve the parents and
71 interfaces for interface, classe and struct
74 * Semantic analysis: since C# can not resolve in a
75 top-down pass what identifiers actually mean, we
76 have to postpone this decision until the above steps
79 * Code generation: The code generation is done through
80 the System.Reflection.Emit API.
85 The compiler performs a number of simple optimizations on its input:
86 constant folding (this is required by the C# language spec) and
87 can perform dead code elimination.
89 Other more interesting optimizations like hoisting are not possible
90 at this point since the compiler output at this point does not
91 generate an intermediate representation that is suitable to
92 perform basic block computation.
94 Adding an intermediate layer to enable the basic block
95 computation to the compiler should be a simple task, but we
96 are considering having a generic CIL optimizer. Since all the
97 information that is required to perform basic block-based
98 optimizations is available at the CIL level, we might just skip
99 this step altogether and have just a generic IL optimizer that
100 would perform hoisting on arbitrary CIL programs, not only
101 those produced by MCS.
103 If this tool is further expanded to perform constant folding
104 (not needed for our C# compiler, as it is already in there)
105 and dead code elimination, other compiler authors might be
106 able to use this generic CIL optimizer in their projects
107 reducing their time to develop a production compiler.
111 MCS was able to parse itself on April 2001, MCS compiled itself
112 for the first time on December 28 2001. MCS became self hosting
113 on January 3rd, 2002.
115 The Mono Runtime and the Mono execution engine were able to make
116 our compiler self hosting on March 12, 2002.
118 ** Questions and Answers
120 Q: Why not write a C# front-end for GCC?
122 A: I wanted to learn about C#, and this was an exercise in this
123 task. The resulting compiler is highly object-oriented, which has
124 lead to a very nice, easy to follow and simple implementation of
127 I found that the design of this compiler is very similar to
128 Guavac's implementation.
130 Targeting the CIL/MSIL byte codes would require to re-architecting
131 GCC, as GCC is mostly designed to be used for register machines.
133 The GCC Java engine that generates Java byte codes cheats: it does
134 not use the GCC backend; it has a special backend just for Java, so
135 you can not really generate Java bytecodes from the other languages
138 Q: If your C# compiler is written in C#, how do you plan on getting
139 this working on a non-Microsoft environment.
141 We will do this through an implementation of the CLI Virtual
142 Execution System for Unix (our JIT engine).
144 Our JIT engine is working for the purposes of using the compiler.
145 The supporting class libraries are being worked on to fully support
150 A: No, currently I am using Jay which is a port of Berkeley Yacc to
151 Java that I later ported to C#. This means that error recovery is
152 not as nice as I would like to, and for some reason error
153 productions are not being caught.
155 In the future I want to port one of the Bison/Java ports to C# for
158 Q: Should someone work on a GCC front-end to C#?
160 A: I would love if someone does, and we would love to help anyone that
161 takes on that task, but we do not have the time or expertise to
162 build a C# compiler with the GCC engine. I find it a lot more fun
163 personally to work on C# on a C# compiler, which has an intrinsic
166 We can provide help and assistance to anyone who would like to work
169 Q: Should someone make a GCC backend that will generate CIL images?
171 A: I would love to see a backend to GCC that generates CIL images. It
172 would provide a ton of free compilers that would generate CIL
173 code. This is something that people would want to look into
174 anyways for Windows interoperation in the future.
176 Again, we would love to provide help and assistance to anyone
177 interested in working in such a project.
179 Q: What about making a front-end to GCC that takes CIL images and
180 generates native code?
182 A: I would love to see this, specially since GCC supports this same
183 feature for Java Byte Codes. You could use the metadata library
184 from Mono to read the byte codes (ie, this would be your
185 "front-end") and generate the trees that get passed to the
188 Ideally our implementation of the CLI will be available as a shared
189 library that could be linked with your application as its runtime
192 Again, we would love to provide help and assistance to anyone
193 interested in working in such a project.
195 Q: But would this work around the GPL in the GCC compiler and allow
196 people to work on non-free front-ends?
198 A: People can already do this by targeting the JVM byte codes (there
199 are about 130 compilers for various languages that target the JVM).
201 Q: Why are you writing a JIT engine instead of a front-end to GCC?
203 A: The JIT engine and runtime engine will be able to execute CIL
204 executables generated on Windows.
206 You might also want to look at the <a href="faq.html#gcc">GCC</a>
207 section on the main FAQ