This was bugging me after I saw all the strings; it feels like this is why we have EnumBase on the toolchain side.
I've included the move of EnumBase to //common because I figured it's reasonable to evaluate together; if we don't want EnumBase in this case, it doesn't make sense to move.
This starts handling return types on functions, and comparing types with `return` statements.
Note, errors remain poor because the type literal is currently associated with a builtin, losing the parse_node that specified it. This means we don't have the original source location to associate with, even though it may be helpful to point at the type in source. We could point at the signature overall, but my leaning is that we wouldn't want that long-term, so TODOs for now and may want to change a little about how the parse node is tracked once things are a little further along.
For parameters (and in the future, arguments too; generally comma-separated lists) track two node blocks:
1. param_ir: The complete IR.
2. param_refs: Nodes within the IR that are the "root" parameter.
param_refs should allow quick counting of the # of parameters, and more efficient comparison of call args with function parameters. param_ir should be necessary to generate the actual signature.
In order to construct this, this refactors the node_block_stack into its own class, which is reused in params_stack. These carry references to the underlying SmallVector for lazy modification in order to avoid a dependency cycle with SemanticsIR (also see notes on empty node blocks below).
When finalized, the block pair is pushed onto finished_params_stack. That's because node_stack only has space for one thing, and this is two things -- so I'm essentially choosing a trade-off of adding another stack in order to avoid consuming more space in the expectation that most parse nodes have 0 or 1 things to return, and 2 will be very rare.
As factored, this currently consolidates most empty node blocks into a single canonical empty node block. This is because I think empty blocks, i.e. `()`, will be very common. In order to achieve this, SemanticsNodeBlockStack does lazy creation.
An alternative approach would have been to use 1 node block per parameter. We decided against this in order to reduce the number of vectors being created.
The parser and lexer already produce YAML, so this is fundamentally a consistency issue. I've been thinking about this, and was looking again because I'm working on adding callables, and figured I'd just fix it now.
Currently, there's a mix of accessing node_stack_ both directly and indirectly, and there are already several Push/Pop methods to help wrap the behavior. However, there's also direct access because of shifting over time, as well as variations in _how_ the stack is used.
This migrates to a separate class in order to make a more specific contract for the API. It cleans up existing uses and adds APIs where needed.
The goal here is to (significantly) reduce the boilerplate needed when defining classes that wrap enums, especially those managed with the `.def`-file style X-macros that are common in the toolchain.
This should also provide both better and more consistent functionality to those classes once ported over to it.
Initially, only `ParserState`, `SemanticsNodeKind`, and `SemanticsBuiltinKind` are ported as these were also the three that JonMeow ported in his original pull/2453 "option 5". This is heavily based on that version of the code.
Goals I was considering that influenced the design:
- Keep the individual enum-wrapping classes as simple and easy to read as possible. Especially important is keeping the `.def` files that are often filled with really important documentation clean and easy to maintain over time.
- Don't rely on computed `#include`s as that is an especially dark corner of the preprocessor and breaks some build systems.
- Have a really good API of the enum-wrapping class, including nice constant names for the values, easy printing, and even easy debugger-callable methods to get the name (as opposed to the integer value).
- Keep the API that users interact with in the base class as clean and easy to read as possible.
- Reduce the boiler plate for each instance of these as much as possible.
- Avoid excessive inline generated code or constants that would result in steady growth in object file sizes and linker effort doing deduplication.
These goals aren't always compatible, so we end up needing to pick a compromise between them when in tension. I think this version is a pretty good compromise.
The original version I started with already pull most of the API into a CRTP-style base class. This version pulls *all* of the common API. This is the main tool for getting consistency and avoiding duplication. However, connecting this base class to the individual enum wrappers is still difficult. Some specific changes here that try to do as much as possible there:
- Use a slightly fancier macro pattern to reduce the boilerplate of defining the raw `enum class` prior to the wrapper class.
- Use a macro to simplify naming the base class.
- Move the name table to a `.cpp` file to avoid every inclusion generating a complete copy of the strings (that the linker has to deduplicate). This is done with some care to sharply reduce the boilerplate needed in that `.cpp` file.
- Sink the name _API_ fully into the CRTP base class. This requires some significant complexity in the implementation, but all of that is hidden behind a single implementation detail macro, and the API itself is simple and readable. This also makes it much more reasonable to test the entire system a single time next to the base class.
This version also moves from constant factory functions to normal constants. This requires two batches -- first a declaration, and then a definition -- but the API result is significantly better and similar to the original option, the macro structure reduces the cost of these. Unfortunately that makes the adoption a bit noisy, but I think its worth the churn.
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
This is a first pass at what semantic type checking might look like. Types propagate along nodes, we use an InvalidType object when there's an error, and once there's an InvalidType we stop doing so much type checking.
This adds some RealLiteral handling in order to get type mismatches. I'm cautious about creating some real value for SemanticsIR (since the tokenized buffer version is a bit constrained), so I'm not doing that yet. But I will probably need to in order to maintain SemanticsIR having hermetic copies of its data, without a parse tree dependency.
This starts adding builtins with TypeType and IntegerLiteralType. Note, structurally that's all they are, and not directly accessible in any way.
Adds a type field to SemanticsNode. Now, IntegerLiteralType can be identified as having type=TypeType, and IntegerLiteral as type=IntegerLiteralType. The current iteration doesn't do anything for type propagation, because I wanted to avoid making this too big.
This also switches the Identifier IR to instead BindName, with some side-effects. I'd been trying to think how to provide a name for TypeType, and switching around how things worked seemed like a better approach. And while I think it's the right direction (e.g., alias should just be a BindName), I also realized I don't need to name TypeType: there's probably a keyword to refer to the builtin, so it shouldn't use regular name lookup.
I'm thinking about how to handle multiple files, and I think the current IRFactory is useful as a file-focused thing. So shifting/renaming accordingly. (doing this in its own PR to make the history a little cleaner for git's move detection)
This rewrites semantics towards a more pure instruction model, in pursuit of the simple instruction-style output.
I think I can get this approach to type-check as it goes along, but obviously this change doesn't prove that yet. I'm separating it out because it's a large rewrite of the semantics structure, tossing out a lot of what was there before. But I think it does help towards several requests, like setting up a clear path for consolidating duplicate identifiers and making the node style more standardized.
I expect to need to pass multiple args to function calls, that'd probably be storing vectors of args similar to how I'm showing identifiers and integer literals stored.
This removes the semantics namespace because (a) it was getting annoying writing the `::` everywhere, and (b) I think the leaning with Carbon is to avoid namespaces (@chandlerc asked not to put SemanticsIR/SemanticsFactory in a namespace, which is the crux of the issue). But, it's still necessary to avoid name conflicts so I just prefix everything with "Semantics" (still a lot of typing, but no `::`).
In summary - some of the changes here are focused on producing the same test results for SemanticsIR, but I think the next step will be to change the SemanticsIR structure to reduce how much is added to the traversal stack.
Switching semantics to a postorder traversal is intended to be more efficient. The traversal stack is to eliminate risk of recursion limits within the semantic analysis that could come from layered code structures. However, we need to start considering the implications for type-checking and what the ParseTree looks like, as well as copying of data here.
As we start thinking about type-checking in SemanticsIR, it's helpful for a function to know its own signature in order to perform lookup recursive calls. The challenge in the post-order walk without this change is it doesn't know it's in a function definition (or similar) until it reaches the FunctionDeclaration; this restructures so that either:
1. For a declaration, the signature is a child of FunctionDeclaration(";")
2. For a definition, the signature is a child of FunctionDefinitionStart("{") which pairs with FunctionDefinition("}"), replacing CodeBlock.
This similarly reorients CodeBlock to be CodeBlockStart("{") as the first child of CodeBlock("}"). I'm not doing that with ParameterList here just because it affects a bit more, and felt like it could be delayed.
Overall, my goal is making the postorder traversal more intuitive along scope boundaries. I think we may also not need subtree_size, so I'm avoiding use of that now.
Currently the SemanticsIRFactory implementation is less clean than I might like (there are a couple comments to this point), but I was starting to feel like a more complete rewrite would be appropriate rather than trying to clean it up further: in particular, I think the node structures are off, but changing them is significant and also changes test output; in turn it may also warrant more substantial ParseTree changes. If you prefer from a reviewer POV, I can do a more complete rewrite.
This builds on semantics-ir lit support added by #2222
The googletest setup was feeling cumbersome, especially as I'm thinking about how to add more testing: I feel like I'm wrestling with the infrastructure.
The `[[ID1]]` and so on in tests is one advantage of switching: it's easier to do matching of IDs for verification. This is also more agnostic about the numbers than before, something which I'm concerned will be important as I think about builtins.
To explain my builtins thought, I think that needs to be another SemanticsIR with basically names pointing at builtin things. But this (a) creates multiple SemanticsIRs, which would confuse the current singleton approach and (b) starts creating more fluctuation for IDs, potentially impacting the numbers used (also, chandlerc's suggested pointers for some use-cases).
Overall it felt like I was heading towards a situation with googletest where writing the tests would be really difficult, and it was adding to my hesitance to write more code in the toolchain. I'm hoping this acts as a simplification.
Note, the "cp" commit has some incremental changes to googletest that I'd considered for making it easier to add matchers, but ultimately I ended up with this outcome.
This is how I'm interpreting discussion:
- Basic elements are getting set to an ID.
- SetName exists to assign a name (which can then be referred to later with an identifier expression) to an ID.
- Expressions are broken down into a series of operations which operate on IDs.
So with something like the last test:
```
fn Main() { return 12 + 34; }
```
This becomes:
```
Function(%0,
{IntegerLiteral(%3, 12),
IntegerLiteral(%2, 34),
BinaryOperator(%1, +, %3, %2),
Return(%1),
})
SetName(`Main`, %0)
```
Note I'm treating blocks as fairly equal to the top of a file now, and basically eliminating boundaries between things. That's because we have discussed also supporting code like:
```
fn Foo() {
fn Bar() {}
Bar();
}
```
Here a declaration of a function is occurring inside a code block, so it felt like eliminating the difference was the best choice.
I know you'd commented on the separation of nodes to individual files before; I still think we're going to have a lot of different types of nodes, and so separating them out into individual files makes them easier to browse.
Working on toolchain semantics:
- SemanticsIR is set up as a container for the semantic tree.
- SemanticsIRFactory builds the tree, with separate transformations for each ParseNodeKind.
- ParseSubtreeConsumer is a helper for transforming a ParseTree::Node's children, managing size/nodes to prevent errors.
- The nodes subdirectory contains SemanticIR nodes.
- MetaNode is used to represent nodes which have "sub-classes": Statements, Declarations, and Expressions.
- MetaNodeBlock is used to represent nodes which exist together in a block with name lookup: Statements and Declarations (not Expressions).
This is traversing children first in order to address the RPO format of ParseTree. This means that when lists are formed, they're reversed to be in code-order (`FixReverseOrdering`).
This is still very much incomplete -- the main intent at present is to demonstrate structure.