-* Optimizations
+===========================================
- Handle if (!x) converting to remove the `!' and instead of using
- a brfalse use a brtrue to jump to the end.
+* Value Parameter
-* LValueResolve
+ I believe that `Value Parameter' might have been introduced
+ after C# 1.0, also notice than in the treatment of Value Parameter
+ the parameters are defined in four categories:
- Maybe we should only call Resolve on RValues and LValueRsolve
- on LValues so that DoREsolve can flag errors on properties
- missing `get' and indexers without a get indexer.
+ Section 9.3 in the latest spec.
-* MemberLookup
- Should take a Location, and pass this information down, so we
- can report better errors.
+Large project:
+--------------
-* Emitcontext
+New
+---
- Do we really need to instanciate this variable all the time?
+ It would be nice to optimize the case of:
- It could be static for all we care, and just use it for making
- sure that there are no recursive invocations on it.
+ Method (new ValueType ())
-* Static-ization
+ So that no temporary is created, and we only use a newobj call
+ that remains on the stack, as opposed to ldloca, initobj, ldloc
+ call.
- Since AppDomain exists, maybe we can get rid of all the stuff
- that is part of the `compiler instance' and just use globals
- everywhere.
+NEW NOTES:
+----------
-* Reorganization: calling structure methods
+ ImplicitStandardConversionExists and ImplicitStandardConversion
+ should always be the same, but there are a few diverging lines that
+ must be studied:
- Structure methods do not expect a `this' pointer, they expect
- an address on the stack. We can probably use all the LValue members
- to do this.
+ if (expr_type == target_type && !(expr is NullLiteral))
+ return expr;
- Now, for cases like:
+ vs:
- struct A { void M ();}
- class B { A Func () {}}
+ if (expr_type == target_type)
+ return true;
- When we do:
- b.Func ().M ()
+****************************************************************************************
+*
+* The information on the rest of this file is mostly outdated, and its kept here for
+* historical reasons
+*
+****************************************************************************************
+
+Error Reporting:
+----------------
- A temporary is created by the compiler to store the result from
- b.Func, we could then use this LocalVariable to invoke the
- AddressOf on it.
+ * Make yyerror show a nice syntax error, instead of the current mess.
-* FindMembers
+Optimization ideas
+------------------
- Move our utility FindMembers from TypeContainer to Decl, because interfaces
- are also scanned with it.
+ Currently when we build a type cache, it contains private members,
+ internal members, and internal protected members; We should trim
+ these out, as it shows up on the profile.
-* Ordering
+ We create too many Arraylists; When we know the size, we should create
+ an array;
- Can a constant_expression invoke overloaded operators?
- Explicit user-defined conversions?
+ During parsing we use arraylists to accumulate data, like this:
-* Visibility
+ thing:
+
+ thing_list
+ : thing { $$ =new ArrayList (); $$.Add ($1); }
+ | thing_list thing { ArrayList a = $1; a.Add ($2); $$ = a; }
- I am not reporting errors on visibility yet.
+ We probably could start using "Pairs" there:
-* Error handling
+ thing_list
+ : thing { $$ = new Pair ($1, null); }
+ | thing_list thing { Pair p = $1; $$ = new Pair ($2, $1); }
- Normalize, and use Tokenizer location
-* Enumerations
+EmitContext.ResolveTypeTree
+---------------------------
- Currently I am not resolving enumerations.
+ We should investigate its usage. The problem is that by default
+ this will be set when calling FindType, that triggers a more expensive
+ lookup.
- Either I track them with `RecordEnum' as I do with classes,
- structs and interfaces or I rewrite the code to visit type
- containers and `walk' the enums with this process.
+ I believe we should pass the current EmitContext (which has this turned off
+ by default) to ResolveType/REsolveTypeExpr and then have the routines that
+ need ResolveType to pass null as the emit context.
-* Known problems:
+DeclareLocal audit
+------------------
- Cast expressions
+ DeclareLocal is used in various statements. The audit should be done
+ in two steps:
- They should should use:
+ * Identify all the declare locals.
- OPEN_PARENS type CLOSE_PARENS
+ * Identify its uses.
- instead of the current production which is wrong, because it
- only handles a few cases.
+ * Find if we can make wrapper functions for all of them.
- Complex casts like:
+ Then we can move DeclareLocal into a helper class.
- Array r = (string []) object
+ This is required to fix foreach in iterators.
- Wont be parsed.
-
-* Interfaces
+Ideas:
+------
- For indexers, the output of ix2.cs is different from our
- compiler and theirs. They use a DefaultMemberAttribute, which
- I have yet to figure out:
+ Instead of the hack that *knows* about System.Object not having any children classes,
+ we should just make it simple for a probe to know that there is no need for it.
- .class interface private abstract auto ansi INTERFACE
- {
- .custom instance void [mscorlib]System.Reflection.DefaultMemberAttribute::.ctor(string)
- = ( 01 00 04 49 74 65 6D 00 00 ) // ...Item..
- ...
- }
+Dead Code Elimination bugs:
+---------------------------
-* Interface indexers
+ I should also resolve all the children expressions in Switch, Fixed, Using.
- I have not figured out why the Microsoft version puts an
- `instance' attribute, and I am not generating this `instance' attribute.
+Major tasks:
+------------
+ Properties and 17.6.3: Finish it.
-* Constructors
+readonly variables and ref/out
+
+BUGS
+----
- Currently it calls the parent constructor before initializing fields.
- It should do it the other way around.
+* Break/Continue statements
-* Use of EmitBranchable
+ A finally block should reset the InLoop/LoopBegin/LoopEnd, as
+ they are logically outside the scope of the loop.
- Currently I use brfalse/brtrue in the code for statements, instead of
- using the EmitBranchable function that lives in Binary
+* Break/continue part 2.
-* Create an UnimplementedExpcetion
+ They should transfer control to the finally block if inside a try/catch
+ block.
- And use that instead of plain Exceptions.
+*
+> // CSC sets beforefieldinit
+> class X {
+> // .cctor will be generated by compiler
+> public static readonly object O = new System.Object ();
+> public static void Main () {}
+> }
+>
-* ConvertImplicit
+PENDING TASKS
+-------------
- Currently ConvertImplicit will not catch things like:
+* Merge test 89 and test-34
- - IntLiteral in a float context to generate a -FloatLiteral.
- Instead it will perform an integer load followed by a conversion.
+* Code cleanup
-* In class.cs: Method.Define
+ The information when registering a method in InternalParameters
+ is duplicated, you can always get the types from the InternalParameters
- Need to use FindMembers to lookup the member for reporting
- whether a new is needed or not.
+* Emit modreq for volatiles
-* virtual-method.cs breaks
+ Handle modreq from public apis.
- It breaks on the call to: new B ();
+* Merge tree.cs, rootcontext.cs
- Where B is a class defined in the source code, my guess is that
- the look for ".ctor" fails
+OPTIMIZATIONS
+-------------
-* Foreach on structure returns does not work
+* User Defined Conversions is doing way too many calls to do union sets that are not needed
- I am generating invalid code instead of calling ldarga for the
- structure, I am calling ldarg:
+* Add test case for destructors
- struct X {
- public IEnumerator GetEnumerator ();
- }
+* Places that use `Ldelema' are basically places where I will be
+ initializing a value type. I could apply an optimization to
+ disable the implicit local temporary from being created (by using
+ the method in New).
- X x;
+* Dropping TypeContainer as an argument to EmitContext
- foreach (object a in x){
- ...
- }
+ My theory is that I can get rid of the TypeBuilder completely from
+ the EmitContext, and have typecasts where it is used (from
+ DeclSpace to where it matters).
+
+ The only pending problem is that the code that implements Aliases
+ is on TypeContainer, and probably should go in DeclSpace.
+
+* Tests
+
+ Write tests for the various reference conversions. We have
+ test for all the numeric conversions.
- I need to get the address of that bad boy
+* Optimizations: variable allocation.
-* Work todo:
+ When local variables of a type are required, we should request
+ the variable and later release it when we are done, so that
+ the same local variable slot can be reused later on.
- Understand how methods are invoked on structs
+* Add a cache for the various GetArrayMethod operations.
-* Using Alias
+* MakeUnionSet Callers
- Need to reset the aliases for each compilation unit, so an
- alias defined in a file does not have any effect on another one:
+ If the types are the same, there is no need to compute the unionset,
+ we can just use the list from one of the types.
- File.cs
- =======
- namespace A {
- using X = Blah;
+* Factor the lookup code for class declarations an interfaces
+ (interface.cs:GetInterfaceByName)
- class Z : X { <-- This X is `Blah'
+RECOMMENDATIONS
+---------------
+
+* Use of lexer.Location in the parser
+
+ Currently we do:
+
+ TOKEN nt TERMINAL nt TERMINAL nt3 {
+ $$ = new Blah ($2, $4, $6, lexer.Location);
}
- File2.cs
- namespace {
- class Y : X { <-- This X Is not `Blah'
- }
+ This is bad, because the lexer.Location is for the last item in `nt3'
+
+ We need to change that to use this pattern:
+
+ TOKEN { oob_stack.Push (lexer.Location) } nt TERMINAL nt TERMINAL nt3 {
+ $$ = new Blah ($3, $5, $7, (Location) oob_stack.Pop ());
}
- I think we can implement Aliases by having an `Alias' context in all
- the toplevel TypeContainers of a compilation unit. The children typecontainers
- just chain to the parents to resolve the information.
+ Notice how numbering of the arguments changes as the
+ { oob_stack.Push (lexer.Location) } takes a "slot" in the productions.
- The driver advances the Alias for each file compiled, so that each file
- has its own alias set.