ILSpy

Commit Graph

Author	SHA1	Message	Date
Siegfried Pammer	d41032efbe	Bind instance compound assignment operators when building expressions "x op= y" and a prefix increment resolve in two phases: the instance operators reachable from the static type of x come first wherever x is a variable - whether or not the result is used - and the static operators only if none of them is applicable. A postfix increment whose result is used is the one form that always binds a static operator. CallBuilder models the first phase when it checks that recompiling an instance operator call binds the same method; the fallback phase is deliberately not modeled, since a call the instance candidates cannot account for must give up the operator form rather than collide with a static operator. The same rule cuts the other way for the folds built from static operator calls: a shadowed fold has to be written in one of the static-binding forms. The binary operators become "x = x + y"; an increment becomes a postfix increment whose result goes to a discard, "_ = x++;". That form only exists as a statement, so the folds that would embed a shadowed increment in an expression hold back and FixRemainingIncrements gives the increment a statement of its own instead. A foreach variable cannot be such a receiver at all (it is read-only), so the loop keeps the existing variable as a writable copy, both for a plain loop variable and for a deconstruction. Assisted-by: Claude:claude-opus-5:Claude Code	3 weeks ago
Siegfried Pammer	04505bbdf2	Highlight and scroll to the changed instruction in the ILAst view The C# debug-steps view highlights and centers the exact AST node a transform changed; the ILAst view already had the step tree and replay-at-step but produced no highlight. Bring it to parity. IL rendering has no token-writer seam like the C# output visitor, so per-instruction text spans are recorded by bracketing ILInstruction.WriteTo via a new INodeTrackingOutput. The dominant inst.ReplaceWith(newInst) transform pattern detaches the instruction passed to Step, so ILTransformContext gains EndStep to record the produced instruction; Stepper additionally records the position's ancestor chain as fallback candidates before the step-limit throw, so the "show state before" view -- which halts at the selected step -- still resolves to a surviving ancestor (ultimately the ILFunction). The highlight-range resolver is shared with the C# language. Assisted-by: Claude:claude-opus-4-8:Claude Code	3 months ago
Siegfried Pammer	77d9ee73e9	Fix IDE2000 globally	9 months ago
ds5678	15223daabd	WIP preincrement improvements	1 year ago
Siegfried Pammer	7b1f8a305c	Implemented support for string concatenation compound assignments involving ReadOnlySpan<char>.	2 years ago
Siegfried Pammer	d5cec0ab37	Fix #3103 : unaligned.stobj cannot be transformed into inline assignment	3 years ago
Daniel Grunwald	b80d20b15d	Add support for calling `operator checked`.	3 years ago
Daniel Grunwald	3c46271a11	Add support for unsigned right shift. Only for user-defined operators so far; builtin right shifts still cast to the appropriate type.	3 years ago
Daniel Grunwald	f04acbdd74	Simplify `IsImplicitTruncation` by using an enum with 3 options instead of a pair of bools. This also fixes the logic for combining the results for BinaryNumericInstruction/IfInstruction.	3 years ago
ElektroKill	ea1f6e7c25	Fixed compound assignment and post/pre increment for pointer dereference	3 years ago
ElektroKill	25873a68e0	Improve support for post-increment/decrement on pointers	4 years ago
Siegfried Pammer	c5ff0cafc5	TransformInlineAssignmentStObjOrCall: Make sure no dead store is produced by the transform.	4 years ago
Daniel Grunwald	bd9aabeae2	Update to dotnet-format 5.1.225507. This makes the formatting expected by our CI consistent with that generated by VS2019.9.	5 years ago
Siegfried Pammer	d2fc380e0e	Fix #2349 : Use proper ExpressionType with DynamicCompoundAssign.	6 years ago
dotnet format	0d9f871a4f	#2128 : Reformat the whole code base.	6 years ago
Daniel Grunwald	1d684102ac	Support compound assignment for native integers.	6 years ago
Daniel Grunwald	93806b46fa	Fix compound assignments with type `bool`. This required removing the "Replace bit.and with logic.and" transform, as it interfered with the compound assignment transform.	7 years ago
Chicken-Bones	661acdfc08	Transform post-increment on float/double. Fixes #1764	7 years ago
Daniel Grunwald	752f0de978	Further adjustments to transforms now that copy propagation no longer runs.	7 years ago
Daniel Grunwald	97d490a3df	Slightly generalize the post-increment transform so that it no longer depends on copy propagation.	7 years ago
Daniel Grunwald	539e3a906d	Fix #1552 : user-defined and decimal increments For user-defined increments, there were problems with Roslyn was optimizing out one of the stores. The new transform FixRemainingIncrements now takes increments/decrements that were not detected by TransformAssignment and introduces a temporary variable that can be incremented. This sometimes requires un-inlining via the new ILInstruction.Extract() operation. Extract() is not supported in all possible contexts, so it is possible but unlikely that some op_Increment calls remain. For decimals, the situation is different: legacy csc actually was optimizing "d + 1m" to "op_Increment(d)", so we can get rid of any left-over increments by undoing this optimization. This now happens in ReplaceMethodCallsWithOperators.	7 years ago
Daniel Grunwald	33c7425fa2	Fix several bugs with compound assignments on indexers. Closes #1580.	7 years ago
Daniel Grunwald	b32cae412a	Eliminate BlockKind.PostfixOperator. We can now represent all compound assignments using the CompoundAssignInstruction, including those on local variables. Closes #1583.	7 years ago
Daniel Grunwald	f2151972a0	Redesign CompoundAssignmentInstruction to support an 'Address' mode where we don't implicitly turn an LdObj into an StObj, but instead directly operate on an address. This means we can get rid of the special case in TransformDisplayClassUsage, as compound.assign can now also be used with the address of a local variable.	7 years ago
Daniel Grunwald	a6def4cdf5	Fix incorrect type for numeric.compound.assign when the pointer type is incompatible with the store type. Closes #1511, #1530, #1533.	7 years ago
Siegfried Pammer	00d6c6175a	#1049 : Check decompiler settings in TransformAssignment.HandleCompoundAssign as well.	8 years ago
Siegfried Pammer	087cb2489d	Fix #1049 : Unimplemented decompiler settings	8 years ago
Siegfried Pammer	c1fca21e8a	Make ILRange field private - introduce public API for IL range manipulation.	8 years ago
Siegfried Pammer	667e04462a	Fix #1325 : Do not perform inline assignment transform on setter calls, that belong to parameterized properties.	8 years ago
Daniel Grunwald	12e74daf48	Handle Comp in InferType().	8 years ago
Daniel Grunwald	ba0a3af3c4	#1195 : Fix several issues decompiling SharpSvn.dll (C++/CLI)	8 years ago
Daniel Grunwald	066da1ea16	Fix #1200 : assertion in TransformInlineAssignmentLocal when reconstructed stack slot (after async-await transform) has small integer type. The assertion was unnecessary, because we already check 'IsImplicitTruncation(inst.Value, inst.Variable.Type)'.	8 years ago
Daniel Grunwald	4580eab7ab	Disable ref inline assignments. This increases the chances that we'll avoid CS8174: A declaration of a by-reference variable must have an initializer.	8 years ago
Daniel Grunwald	33ef82f75c	Fix #1082 : Add support for compound assignment with string.Concat(). Currently only supports the binary overloads of string.Concat().	8 years ago
Siegfried Pammer	c4f41f459f	Add support for more dynamic binary operators.	8 years ago
Siegfried Pammer	273a1c24ff	Add dynamic compound assignment transforms	8 years ago
Daniel Grunwald	f021ec4383	NumericCompoundAssign: implement ILiftableInstruction	8 years ago
Daniel Grunwald	3956fa685d	Refactor TransformAssignment to reduce code duplication.	8 years ago
Siegfried Pammer	9ebfbe57dd	Add transforms for userdef compound assignment fix bugs in other transforms	8 years ago
Daniel Grunwald	25706f2a39	Support logic.and/etc. in IsImplicitTruncation().	8 years ago
Daniel Grunwald	75a627d40b	Don't use compound assignment when the RHS value does not fit into the LHS type.	8 years ago
Daniel Grunwald	0a1921ff67	Fix #1007 : post-increment transform incorrect due to variable splitting.	9 years ago
Siegfried Pammer	349183544a	Rename BlockType to BlockKind	9 years ago
Daniel Grunwald	a82430fc78	Add missing safety checks to inline assignment transforms.	9 years ago
Daniel Grunwald	11631d6cf0	Remove redundant code from TransformAssignment.	9 years ago
Daniel Grunwald	7847a69d69	Fix compound assignment and post-increment.	9 years ago
Daniel Grunwald	4c5f0b7e9c	Convert TransformAssignment into a statement transform and add support for inline property assignments.	9 years ago
Daniel Grunwald	40a4f08a8d	Add support for compound assignment of short integers.	9 years ago
Daniel Grunwald	248a9ad76f	Fix #954 : don't use invalid compound assignments on enums	9 years ago
Daniel Grunwald	b0bf6326ad	Allow inline assignments to short integers when we can tell that there's no truncation.	9 years ago

1 2

69 Commits (826cf02254f4dcebd3e0097a2930d10998dd29af)