1 The Internals of the Mono C# Compiler
9 The Mono C# compiler is a C# compiler written in C# itself.
10 Its goals are to provide a free and alternate implementation
11 of the C# language. The Mono C# compiler generates ECMA CIL
12 images through the use of the System.Reflection.Emit API which
13 enable the compiler to be platform independent.
15 * Overview: How the compiler fits together
17 The compilation process is managed by the compiler driver (it
20 The compiler reads a set of C# source code files, and parses
21 them. Any assemblies or modules that the user might want to
22 use with his project are loaded after parsing is done.
24 Once all the files have been parsed, the type hierarchy is
25 resolved. First interfaces are resolved, then types and
28 Once the type hierarchy is resolved, every type is populated:
29 fields, methods, indexers, properties, events and delegates
30 are entered into the type system.
32 At this point the program skeleton has been completed. The
33 next process is to actually emit the code for each of the
34 executable methods. The compiler drives this from
37 Each type then has to populate its methods: populating a
38 method requires creating a structure that is used as the state
39 of the block being emitted (this is the EmitContext class) and
40 then generating code for the topmost statement (the Block).
42 Code generation has two steps: the first step is the semantic
43 analysis (Resolve method) that resolves any pending tasks, and
44 guarantees that the code is correct. The second phase is the
45 actual code emission. All errors are flagged during in the
48 After all code has been emitted, then the compiler closes all
49 the types (this basically tells the Reflection.Emit library to
50 finish up the types), resources, and definition of the entry
51 point are done at this point, and the output is saved to
54 The following list will give you an idea of where the
55 different pieces of the compiler live:
60 This drives the compilation process: loading of
61 command line options; parsing the inputs files;
62 loading the referenced assemblies; resolving the type
63 hierarchy and emitting the code.
67 The state tracking for code generation.
71 Code to do semantic analysis and emit the attributes
76 Keeps track of the types defined in the source code,
77 as well as the assemblies loaded.
81 This contains the MCS type system.
85 Error and warning reporting methods.
89 Assorted utility functions used by the compiler.
95 The tokenizer for the C# language, it includes also
98 cs-parser.jay, cs-parser.cs:
100 The parser is implemented using a C# port of the Yacc
101 parser. The parser lives in the cs-parser.jay file,
102 and cs-parser.cs is the generated parser.
106 The `location' structure is a compact representation
107 of a file, line, column where a token, or a high-level
108 construct appears. This is used to report errors.
114 Basic expression classes, and interfaces most shared
115 code and static methods are here.
119 Most of the different kinds of expressions classes
124 The assignment expression got its own file.
128 The classes that represent the constant expressions.
132 Literals are constants that have been entered manually
133 in the source code, like `1' or `true'. The compiler
134 needs to tell constants from literals apart during the
135 compilation process, as literals sometimes have some
136 implicit extra conversions defined for them.
140 The constant folder for binary expressions.
146 All of the abstract syntax tree elements for
147 statements live in this file. This also drives the
148 semantic analysis process.
150 Declarations, Classes, Structs, Enumerations
154 This contains the base class for Members and
155 Declaration Spaces. A declaration space introduces
156 new names in types, so classes, structs, delegates and
157 enumerations derive from it.
161 Methods for holding and defining class and struct
162 information, and every member that can be in these
163 (methods, fields, delegates, events, etc).
165 The most interesting type here is the `TypeContainer'
166 which is a derivative of the `DeclSpace'
170 Handles delegate definition and use.
174 Handles enumerations.
178 Holds and defines interfaces. All the code related to
179 interface declaration lives here.
183 During the parsing process, the compiler encapsulates
184 parameters in the Parameter and Parameters classes.
185 These classes provide definition and resolution tools
190 Routines to track pending implementations of abstract
191 methods and interfaces. These are used by the
192 TypeContainer-derived classes to track whether every
193 method required is implemented.
196 * The parsing process
198 All the input files that make up a program need to be read in
199 advance, because C# allows declarations to happen after an
200 entity is used, for example, the following is a valid program:
205 a = "hello"; b = "world";
214 At the time the assignment expression `a = "hello"' is parsed,
215 it is not know whether a is a class field from this class, or
216 its parents, or whether it is a property access or a variable
217 reference. The actual meaning of `a' will not be discvored
218 until the semantic analysis phase.
220 ** The Tokenizer and the pre-processor
222 The tokenizer is contained in the file `cs-tokenizer.cs', and
223 the main entry point is the `token ()' method. The tokenizer
224 implements the `yyParser.yyInput' interface, which is what the
225 Yacc/Jay parser will use when fetching tokens.
227 Token definitions are generated by jay during the compilation
228 process, and those can be references from the tokenizer class
229 with the `Token.' prefix.
231 Each time a token is returned, the location for the token is
232 recorded into the `Location' property, that can be accessed by
233 the parser. The parser retrieves the Location properties as
234 it builds its internal representation to allow the semantic
235 analysis phase to produce error messages that can pin point
236 the location of the problem.
238 Some tokens have values associated with it, for example when
239 the tokenizer encounters a string, it will return a
240 LITERAL_STRING token, and the actual string parsed will be
241 available in the `Value' property of the tokenizer. The same
242 mechanism is used to return integers and floating point
245 C# has a limited pre-processor that allows conditional
246 compilation, but it is not as fully featured as the C
247 pre-processor, and most notably, macros are missing. This
248 makes it simple to implement in very few lines and mesh it
251 The `handle_preprocessing_directive' method in the tokenizer
252 handles all the pre-processing, and it is invoked when the '#'
253 symbol is found as the first token in a line.
255 The state of the pre-processor is contained in a Stack called
256 `ifstack', this state is used to track the if/elif/else/endif
257 nesting and the current state. The state is encoded in the
258 top of the stack as a number of values `TAKING',
259 `TAKEN_BEFORE', `ELSE_SEEN', `PARENT_TAKING'.
263 Locations are encoded as a 32-bit number (the Location
264 struct) that map each input source line to a linear number.
265 As new files are parsed, the Location manager is informed of
266 the new file, to allow it to map back from an int constant to
267 a file + line number.
269 Prior to parsing/tokenizing any source files, the compiler
270 generates a list of all the source files and then reserves the
271 low N bits of the location to hold the source file, where N is
272 large enough to hold at least twice as many source files as were
273 specified on the command line (to allow for a #line in each file).
274 The upper 32-N bits are the line number in that file.
276 The token 0 is reserved for ``anonymous'' locations, ie. if we
277 don't know the location (Location.Null).
279 The tokenizer also tracks the column number for a token, but
280 this is currently not being used or encoded. It could
281 probably be encoded in the low 9 bits, allowing for columns
282 from 1 to 512 to be encoded.
286 The parser is written using Jay, which is a port of Berkeley
287 Yacc to Java, that I later ported to C#.
289 Many people ask why the grammar of the parser does not match
290 exactly the definition in the C# specification. The reason is
291 simple: the grammar in the C# specification is designed to be
292 consumed by humans, and not by a computer program. Before
293 you can feed this grammar to a tool, it needs to be simplified
294 to allow the tool to generate a correct parser for it.
296 In the Mono C# compiler, we use a class for each of the
297 statements and expressions in the C# language. For example,
298 there is a `While' class for the the `while' statement, a
299 `Cast' class to represent a cast expression and so on.
301 There is a Statement class, and an Expression class which are
302 the base classes for statements and expressions.
308 * Internal Representation
312 Expressions in the Mono C# compiler are represented by the
313 `Expression' class. This is an abstract class that particular
314 kinds of expressions have to inherit from and override a few
317 The base Expression class contains two fields: `eclass' which
318 represents the "expression classification" (from the C#
319 specs) and the type of the expression.
321 Expressions have to be resolved before they are can be used.
322 The resolution process is implemented by overriding the
323 `DoResolve' method. The DoResolve method has to set the
324 `eclass' field and the `type', perform all error checking and
325 computations that will be required for code generation at this
328 The return value from DoResolve is an expression. Most of the
329 time an Expression derived class will return itself (return
330 this) when it will handle the emission of the code itself, or
331 it can return a new Expression.
333 For example, the parser will create an "ElementAccess" class
338 During the resolution process, the compiler will know whether
339 this is an array access, or an indexer access. And will
340 return either an ArrayAccess expression or an IndexerAccess
341 expression from DoResolve.
345 *** The Expression Class
347 The utility functions that can be called by all children of
352 Constants in the Mono C# compiler are reprensented by the
353 abstract class `Constant'. Constant is in turn derived from
354 Expression. The base constructor for `Constant' just sets the
355 expression class to be an `ExprClass.Value', Constants are
356 born in a fully resolved state, so the `DoResolve' method
357 only returns a reference to itself.
359 Each Constant should implement the `GetValue' method which
360 returns an object with the actual contents of this constant, a
361 utility virtual method called `AsString' is used to render a
362 diagnostic message. The output of AsString is shown to the
363 developer when an error or a warning is triggered.
365 Constant classes also participate in the constant folding
366 process. Constant folding is invoked by those expressions
367 that can be constant folded invoking the functionality
368 provided by the ConstantFold class (cfold.cs).
370 Each Constant has to implement a number of methods to convert
371 itself into a Constant of a different type. These methods are
372 called `ConvertToXXXX' and they are invoked by the wrapper
373 functions `ToXXXX'. These methods only perform implicit
374 numeric conversions. Explicit conversions are handled by the
375 `Cast' expression class.
377 The `ToXXXX' methods are the entry point, and provide error
378 reporting in case a conversion can not be performed.
382 The C# language requires constant folding to be implemented.
383 Constant folding is hooked up in the Binary.Resolve method.
384 If both sides of a binary expression are constants, then the
385 ConstantFold.BinaryFold routine is invoked.
387 This routine implements all the binary operator rules, it
388 is a mirror of the code that generates code for binary
389 operators, but that has to be evaluated at runtime.
391 If the constants can be folded, then a new constant expression
392 is returned, if not, then the null value is returned (for
393 example, the concatenation of a string constant and a numeric
394 constant is deferred to the runtime).
403 * The semantic analysis
405 Hence, the compiler driver has to parse all the input files.
406 Once all the input files have been parsed, and an internal
407 representation of the input program exists, the following
410 * The interface hierarchy is resolved first.
411 As the interface hierarchy is constructed,
412 TypeBuilder objects are created for each one of
415 * Classes and structure hierarchy is resolved next,
416 TypeBuilder objects are created for them.
418 * Constants and enumerations are resolved.
420 * Method, indexer, properties, delegates and event
421 definitions are now entered into the TypeBuilders.
423 * Elements that contain code are now invoked to
424 perform semantic analysis and code generation.
430 The EmitContext class is created any time that IL code is to
431 be generated (methods, properties, indexers and attributes all
432 create EmitContexts).
434 The EmitContext keeps track of the current namespace and type
435 container. This is used during name resolution.
437 An EmitContext is used by the underlying code generation
438 facilities to track the state of code generation:
440 * The ILGenerator used to generate code for this
443 * The TypeContainer where the code lives, this is used
444 to access the TypeBuilder.
446 * The DeclSpace, this is used to resolve names through
447 RootContext.LookupType in the various statements and
450 Code generation state is also tracked here:
454 This variable tracks the `checked' state of the
455 compilation, it controls whether we should generate
456 code that does overflow checking, or if we generate
457 code that ignores overflows.
459 The default setting comes from the command line
460 option to generate checked or unchecked code plus
461 any source code changes using the checked/unchecked
462 statements or expressions. Contrast this with the
463 ConstantCheckState flag.
467 The constant check state is always set to `true' and
468 cant be changed from the command line. The source
469 code can change this setting with the `checked' and
470 `unchecked' statements and expressions.
474 Whether we are emitting code inside a static or
479 The value that is allowed to be returned or NULL if
480 there is no return type.
485 Points to the Type (extracted from the
486 TypeContainer) that declares this body of code
492 Whether this is generating code for a constructor
496 Tracks the current block being generated.
500 The location where return has to jump to return the
503 A few variables are used to track the state for checking in
504 for loops, or in try/catch statements:
508 Whether we are in a Finally block
512 Whether we are in a Try block
516 Whether we are in a Catch block
519 Whether we are inside an unsafe block
525 Errors are reported during the various stages of the
526 compilation process. The compiler stops its processing if
527 there are errors between the various phases. This simplifies
528 the code, because it is safe to assume always that the data
529 structures that the compiler is operating on are always
532 The error codes in the Mono C# compiler are the same as those
533 found in the Microsoft C# compiler, with a few exceptions
534 (where we report a few more errors, those are documented in
535 mcs/errors/errors.txt). The goal is to reduce confussion to
536 the users, and also to help us track the progress of the
537 compiler in terms of the errors we report.
539 The Report class provides error and warning display functions,
540 and also keeps an error count which is used to stop the
541 compiler between the phases.
543 A couple of debugging tools are available here, and are useful
544 when extending or fixing bugs in the compiler. If the
545 `--fatal' flag is passed to the compiler, the Report.Error
546 routine will throw an exception. This can be used to pinpoint
547 the location of the bug and examine the variables around the
550 Warnings can be turned into errors by using the `--werror'
551 flag to the compiler.
553 The report class also ignores warnings that have been
554 specified on the command line with the `--nowarn' flag.
556 Finally, code in the compiler uses the global variable
557 RootContext.WarningLevel in a few places to decide whether a
558 warning is worth reporting to the user or not.