Chapter 1

Dialect: grouping functionality to extend MLIR system by adding new operators, attributes, and types.
Operations are defined by
- Name
- List of SSA operand values
- List of attributes (MLIR's mechanism for attaching constant metadata to operations)
- List of types for result values
- Source location
- List of successor blocks
- List of regions
Traits: mechanism to inject additional behavior (accessors, verification, etc) into operators.
MLIR allows undefined operations to be manipulated through opaque Operaton class.

Chapter 3: High-level transformations

RewritePatterns are used to represent term rewrites.
Canonicalization framework: used to reduce operations to canonical forms.

Chapter 4: Generic transformations

MLIR declares abstract classes for high level transformations which can be defined on operation specific derived classes.
DialectInlinerInterface

Chapter 5: Partial Lowering

DialectConversion framework is used to convert operations from one dialect to another.
To use the DialectConversion framework, we provide it with two things:
1. A specification of what operations are legal and illegal, used to decide what to convert and what to not.
2. A set of patterns used to convert illegal operations into a set of zero or more legal operations.
DialectConversion framework also uses RewritePatterns, it uses a derived class called ConversionPattern.

TODO

Spelunk adaptor TransposeOpAdaptor
Spelunk helper lowerOpToLoops

Chapter 6: LLVM Lowering

LLVMDialect
TypeConverter