Attach the cleanup to the `Temporary` instruction instead of to the
`TemporaryStorage` instruction. We create `TemporaryStorage`
instructions speculatively when creating an initializing expression, and
may overwrite those instructions with other instructions if it turns out
that a temporary is not required. Instead, wait until we finalize the
temporary and create a `Temporary` instruction to register the cleanup.
We add a virtual node (`CompileTimeBindingPatternStart`) as the first
child of `CompileTimeBindingPattern` which holds the identifier
underneath it, so that it is checked just before the type expression of
the `CompileTimeBindingPattern`. When we reach this virtual node during
check, we add `.Self` as a name in the current scope, and when we reach
`CompileTimeBindingPattern` we remove it from scope, which ensures it's
present during only the checking of the type expression for the compile
time pattern.
At the moment the `.Self` has a different type (it's a `TypeType`) than
other `.Self` in the facet type (which are a single `FacetType`), but
the intention is to immediately substitute it out of the facet type
entirely, replacing it with a reference to the compile time binding (a
`BindSymbolicName`) itself. A TODO has been added for this.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
The self access is important; for the test `generic_class.carbon` being
added to `toolchain/check/testdata/class/destroy_calls.carbon`, it was
using `%T.as.Destroy` instead of `%D.as.Destroy`, indicating the default
blank impl was being used instead of the type-specific version. That
test is trying to focus on the issue, but the delta is visible in a
couple other files in this PR, for example
`toolchain/check/testdata/class/generic/init.carbon`.
I'm separately working on getting rid of the default impl, which is how
I noticed this.
This makes all `.Self` references in a facet type canonically the same
(which will remain true iff they refer to the same `Self` type in the
future), removing the need to do more complex comparisons between them
using the EntityName, interface, and index. This allows the comparison
of types containing `.Self` references to be done correctly regardless
of where the `.Self` appears, as such type expressions will all be
canonically equal if they otherwise equal now, regardless of whether
they are written in the context where `.Self` could have seen different
`Self` facet types.
In order to retain access to constraints on a base `.Self` facet type,
in the case of applying `where` to an existing facet type, we:
- Give the base facet type as a `RequirementBaseFacetType` constraint so
that eval of `WhereExpr` can find and copy all the constraints off of
it.
- Introduce eager/early rewrite constraint resolution, which allows a
constraint to eagerly resolve access to earlier rewrite constraints
(`where .A = () and .B = .A` is eagerly transformed into `where .A = ()
and .B = ()`) before the full constraint resolution step. This allows
use of rewrite constraints in larger type expressions, such as `where .A
= () and .B = C(.A)` and `C` will know that the argument is `()`.
I was thinking about this for destruction, which I may not be able to
use it for, but still think this may be a good change to keep features
consistent.
Although this focused on `Destroy` support, some choices here around
`implicit_type_impls` are because copy/move will likely follow a similar
approach. I'm trying not to predict too much about how we'll structure
those, but I'm putting `Destroy` impl logic in a file that could perhaps
be shared with those. They'd likely be interested in similar things,
e.g. traversing members of types (particularly class, struct literal,
tuple literal).
At present this sets the destroy function as `no_op` which is consistent
with current logic, but has a TODO to correctly define.
Constant importing for functions changes slightly due to some issues I
was having with `GetFunctionType`. zygoloid suggested this approach to
avoid `EvalInst` logic.
Adds a flag for controlling whether to generating these impls. While
this does generation for `class`, as noted above this'll also need to be
done for tuples and struct literals, which would leave the `none.carbon`
min_prelude unable to use any types. Note if destruction *would* occur,
it'll still look up `Core.Destroy` for that and fail, but that's already
true of any test using `none.carbon`. I'm trying to use the flag to see
if we can keep `none.carbon` working mostly-consistently.
I'd tried separating out the flag to #5852, but that got a lot of
pushback over whether the behavior was appropriate. I'm hoping that the
interactions here make it clearer why the particular approach -- the
goal is not to enable advanced testing, or create some new end-user
behavior that we really support, it's just to keep no-prelude tests
functional. The main question raised there was why not just keep
generating `impl T as Core.Destroy` if `fn destroy` is present -- but I
think here it should be apparent that would require additional
complexity, as the generation of `impl T as Core.Destroy` is not
currently conditioned based on the implementation of `fn destroy`. I'd
rather add complexity to this flag only if it's enabling interesting
test functionality.
---------
Co-authored-by: Geoff Romer <gromer@google.com>
Change impls from `<interface>.impl` to `<self>.as.<interface>.impl`,
and *member* functions to `<parent scope>.<fn>` (non-member functions
exclude their parent scope). Stop special-casing builtin functions,
given the new naming scheme.
The purpose of this is to make it clearer when a member function is
being accessed and, if so, which member function. In particular, we
often access interface `Op` functions. The builtin function
special-casing was intended to help with that, but we still have lots of
`Op` functions. This particular approach should make the interactions
clearer.
This changes up queueing of block IDs a little because, in particular,
we need to process bodies of entities only after constants finish
processing. But, it should also result in less memory usage during
processing because it means we have less on the insts stack at any given
time, since we track a block rather than all instructions contained by
the block.
This switches from the `CollectNamesInBlock` approach for entities, to
instead traversing entities as they're encountered. For example, when
traversing constants, when a type is found, the entity will have its
block queued for processing.
This leads to a change in the traversal order, which affects
disambiguation done by numeric sequencing (since that's just showing the
traversal order).
This will allow for simpler "name based on name" logic. This is
something I plan to use for:
- impls: `<type>.as.<interface>.impl`
- functions: `<entity>.<member function>`
- Note an impl may be used as the entity for a bound function.
By naming the entities as they're encountered, I'll be able to rely on
the generated names rather than recalculating them.
To assist this, I'm also differentiating between the ambiguous and
disambiguated name. Otherwise, we could end up with things like
`<function>.<disambiguator>.<call>.<other disambiguator>`, where the
repeated disambiguator may not be necessary in order to get full
disambiguation. It's also a smaller delta from the current output.
Note, changing `Name` to a class felt appropriate given its shape. I was
also noticing that parts of its API were unused, and the class helps
detect unused private members.
---------
Co-authored-by: Geoff Romer <gromer@google.com>
Use `full.carbon` min-prelude for any tests using the full prelude. A
few tests were using it unnecessarily and are changed to a more minimal
one in the process.
Any test which does not include some min-prelude will now fail with an
error.
In `BuildUnaryOperator`, `GetOperatorOpFunction` is treated as
desugaring, but `PerformCompoundMemberAccess` and `PerformCall` are not.
This treats all of destruction as desugaring.
This leads to some instructions being elided, because of `GetOrAddInst`
behaviors:
> // If the instruction has a desugared location and a constant value,
returns
> // the constant value's instruction ID. Otherwise, same as AddInst.
This changes instructions that previously had a non-desugared location
to instead have a desugared location, so if they also have a constant
value then the constant value can be used directly.
This just catches uses equivalent to `return;` and `return <expr>;`.
Note it's just extending the current implicit return logic, not really
adding much unique here.
I'm still delaying break and continue because those require partial
destruction, which is more work and I want to be careful to get it
right.
This changes `Destroy` to use an interface for its implementation.
Note that this change includes a lot of test updates. Even when
`Destroy` is a no-op, it still causes code generation as part of
determining that.
Originally I was trying to use ranges to cut down the scope of this, and
to a degree I think they have. But a flipside here is that cases where
no destructors should be generated -- particularly globals -- would be
needed to completely remove destructor calls. Even for ranges, the range
can often include the destructor placement. So I've shifted
frame-of-thought a little: accept a bunch of destructor churn, because
destructors are needed and will be prevalent. The verbosity is a feature
of the design to make desugaring apparent in IR, not a bug.
This drops file_test runtime from about 3s to 2.5s on my machine, which
is now ~30% faster than before #5653 slowed things down by adding a lot
of stuff to the production prelude.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
If convert fails after applying a substitution in deduce, fail deduction
rather than succeeding deduction with an ErrorInst in the deduced
argument.
For example, in
`toolchain/check/testdata/facet/fail_convert_class_type_to_generic_facet_value.carbon`
the `WrongGenericParam` is deduced for the first argument, then
substituted into the second parameter, but the argument can not convert
to the parameter after substitution. In this case, deduction fails
instead of producig a call with an error in the second argument. The
resulting semir drops the call with an error argument:
```
-// CHECK:STDOUT: %CallGenericMethod.specific_fn: <specific function> = specific_function %CallGenericMethod.ref, @CallGenericMethod(constants.%WrongGenericParam, <error>) [concrete = <error>]
-// CHECK:STDOUT: %CallGenericMethod.call: init %empty_tuple.type = call %CallGenericMethod.specific_fn() [concrete = <error>]
```
The none.carbon min-prelude is not just an empty prelude, it also
prevents any prelude from being imported at all. So no import machinery
runs before the test, only the `package` statement from the prelude
would run.
Use the none.carbon min-prelude in a few tests that were specifying
`--no-prelude-import` to give it a trial run.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Adds min-preludes more tests which were seen as slow and their
surrounding neighbours. This drops the file_test runtime on my machine
from about 7s to about 6s.
For a few files that are clearly only testing diagnostics, we drop the
if-present semir ranges and the associated TODO.
Replace the binary `Operation` interfaces with the `OperationWith(T:!
type)` interfaces described in the design, and add a `Result` associated
type for both unary and binary operations. Update the `impl`s in the
prelude for integer types to use the new form, including supporting
implicit conversion of either operand.
I've tried to split this PR up into commits focused on distinct changes
for review convenience. It may be simplest to review it one commit at a
time.
Also fix substitution into constants to provide a source location. This
matters if the result of substitution ends up being part of a generic
eval block.
---------
Co-authored-by: Dana Jansens <danakj@orodu.net>
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Previously we walked the global variables defined by the current file
and emitted an LLVM global variable definition for each of them. Now
instead, when emitting a constant reference to a global variable, we
emit an LLVM global variable declaration, and we then subsequently walk
the global variables defined by the current file and convert each of
them from a declaration to a definition.
In order to make import of names of global variables work, add support
for import of `var`, as well as support for importing `tuple_access` and
`tuple_pattern` in the case where the `var` has a tuple pattern in its
declaration. Also treat `bind_name`s that are reference bindings to
`var`s as having the same constant reference value as their `var` so
that we can properly import and lower them.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Co-authored-by: Carbon Infra Bot <carbon-external-infra@google.com>
Where `--no-dump-sem-ir` is used, change to `--dump-sem-ir-ranges=only`.
Otherwise, add `--dump-sem-ir-ranges=if-present` with a TODO to change
to `only`.
Note, SemIR is affected just because the extra comments change line
numbers in files where splits aren't in use.
- Track the `VarPattern` instruction on the `VarStorage` instruction so
that it's available for name mangling.
- Mangle global variables based on the first binding name within their
pattern.
- Give global variables external rather than internal linkage, except if
they have no bindings whatsoever in their pattern.
- To support lowering references to bindings nested within a global var,
such as for `var (x: i32, b: i32)`, add some basic initial support for
reference constant expressions. Treat a global `var` as a reference
constant, and treat an aggregate access into a reference constant as a
reference constant.
Instead of building an eval block as a separate pass at the end of a
generic, build the eval block incrementally.
The larger change here is that asking for the type or constant value of
an instruction now always returns an unattached type or constant value,
in order to preserve the behavior that we previously achieved by doing
the rewrite to attached types and constant values at the end of handling
the generic.
This also incidentally fixes some subtle issues where attached types and
constant values would leak out into check and cause it to get confused
about differences between attached and unattached values. Check should
no longer see attached values except where it explicitly asks for them.
---------
Co-authored-by: Dana Jansens <danakj@orodu.net>
Don't import `ImplWitnessTable` into the `constants` block, because we
generally don't put `Unique` constants there. This matches the handling
of the other kinds of `Unique` constants. In order to keep the
instruction visible in formatted SemIR, add it to the `imports` block
instead.
Also fix a bug in the instruction formatter that resulted in
instructions in the `imports` block being omitted from the output if
they were only referenced by earlier instructions in the `imports` block
and by instructions in the `constants` block. This was already resulting
in some referenced instructions being omitted from the output, but also
occurred frequently for `impl_witness_table` instructions after this
change because it is common for the only reference to those instructions
to be from `impl_witness` instructions in the `constants` block.
This was originally needed to support constant evaluation of name
expressions, but that's now done in a different way.
This is actually a step toward treating all patterns as constants. The
upcoming change will do so in a slightly different way, and so it will
simplify the review to start from a baseline where patterns are never
constant.
A `MetaInstId` is intended to represent a handle to an instruction in a
generic as an operand to a template action; substituting into the action
should not substitute into the referenced instruction.
Fixing this exposed a bug in `GetOrAddInst` where it would return an
`InstId` of an unattached symbolic constant in some cases, rather than
the `InstId` of an instruction that has the relevant (attached) constant
value. That's fixed for now by turning off the `GetOrAddInst`
optimization in that case, but in future we can refine this by adding
the instruction to the eval block for the generic only, and not to the
body of the generic.
This allows us to import the table for a given impl only once, while we
can import many ImplWitness instructions with different specifics for a
generic impl.
For example in convert_facet_value_to_narrowed_facet_type.carbon we see
that a single witness table is imported for the BitAnd interface, with
multiple witnesses (for different specifics) imported and sharing the
same table.
The ImplWitnessTable now contains a back-link to the Impl the witness is
for, allowing inst namer to name that interface in the textual semir,
and allowing the interface to be found when debugging from a witness.
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
In preparation for shifting from `TypeId`s potentially representing
attached types to always representing unattached types, using
[terminology suggested on
Discord](https://discord.com/channels/655572317891461132/963846118964350976/1359286326779973712).
This change causes us to track slightly more type spelling information
through SemIR.
One change that has significant impact on the SemIR output is that we
now build a `struct_type` instruction in each class representing the
types of the fields, including the spelling used for those types. This
is now no longer always identical to the corresponding canonical
`struct_type` for the object representation, so it's built separately
and owned by the class.
Also remove `TypeBlock` support entirely, as its only use was
representing `TupleType`s, which now use an `InstBlock`.
This gives a slightly simpler representation for `UnboundElementType`s
in eval blocks, and in principle allows us to preserve the spelling of a
field's type into the `UnboundElementType` and thereby into a field
reference, although as of right now this doesn't affect our diagnostic
output in any way.
During error recovery for a field with a non-concrete type, preserve the
type in the `UnboundElementType` regardless. It's not really problematic
to have a non-concrete type there, and this makes it easier to track the
instruction used to specify the type.
This is a step towards switching symbolic types to always be abstract
during type checking.
For each kind of instruction, specify whether its constant evaluation
needs an `InstId` or not. If it does, ensure that all constant
evaluation of that instruction provides one. Otherwise, allow calling
into the evaluator without providing an `InstId`.
This allows us to reliably use the `InstId` in evaluation steps that
either need a location or need to look at the original operands of the
instruction prior to evaluation, and also to support `TryEvalInst` calls
safely for instructions whose evaluation does not need an `InstId`.
Each of these types takes another type as an operand. Instead of storing
that other type as a `TypeId`, store it as an `InstId` so that we can
track how it was written, not only its canonical form.
The canonical constant values of these types continue to store the
canonical constant values of their operands, as normal.
---------
Co-authored-by: Dana Jansens <danakj@orodu.net>
Emitting definitions in check. This resolves the crash in lowering which
necessitated definitions be emitted.
Some of the test changes need further review.
Factor out logic to evaluate `EntityNameId` instead of duplicating it
between `BindSymbolicName` and `SymbolicBindingPattern`. Remove support
in `SymbolicBindingPattern` for evaluating a pattern to the constant
value of the corresponding binding, which doesn't really make any sense
given that patterns don't generally evaluate to the value that they
matched.
This results in the handling for `SymbolicBindingPattern` being simply
the default handling for an always-constant instruction, so remove the
special case for it entirely and change its constant kind to `Always`.
It's not entirely clear that it makes sense for `SymbolicBindingPattern`
to be treated as a constant when other patterns aren't, but we seem to
be relying on this in various places, so leave it as a constant for now.
Changing it to never be constant will be a smaller change now -- it just
requires changing the `constant_kind`.
The IR changes in the tests are fairly widespread, but mechanical, and
there are two kinds of things changing:
- `symbolic_binding_pattern`s in specifics now evaluate to
`symbolic_binding_pattern`s, not to the argument values. This means in a
few cases we end up with additional `symbolic_binding_pattern`
constants.
- We evaluate the type operand of `symbolic_binding_pattern` now, so an
error in the type will now properly be propagated into an error in the
pattern's constant value.
When transforming instructions with symbolic constant values into the
eval block, we previously special-cased `bind_symbolic_name` (and
`symbolic_binding_pattern`) because they are places where symbolicness
is introduced, rather than propagated from operands, and just copied
them into the eval block. However, `bind_symbolic_name` can be dependent
on other symbolic constants, because it can have a type that is
dependent. In this case, the copy in the eval block would not have its
type properly adjusted to refer to the type within the eval block.
Fix this by performing substitution into `bind_symbolic_name` rather
than copying it directly, and instead, detect cases where substitution
determined that the instruction was unchanged despite having a symbolic
constant value, and force it to be rebuilt in that case.
I've not found any way that the previous behavior actually caused
problems, or affected the observable behavior of the toolchain. The type
of these instructions in the eval block doesn't make much difference to
anything because they get immediately replaced by their corresponding
argument values when we run the eval block. But this came up and caused
some test output churn when I was making a different change, and it
seems like a fix to our representation even if it's not changing
behavior, so I'm splitting it out so it can be handled separately.
When a generic function declaration was encountered for the second or
more time, we would FinishGenericRedecl() for the function decl, but
this just popped the generic region stack and moved on.
The issue with that is when the stack entry is gone, we lose the
symbolic constants from that declaration, and are unable to rewrite them
to point to the actual generic. This left us with a function declaration
with abstract symbolic values that were not useful, and in a function
call we use the declaration attached to the definition, which would be a
declaration with broken symbolic values. Then the function would be
uncallable since deduce would be unable to determine argument types
without the generic bindings.
This resolves the issue for functions, as well as ensuring the correct
generic id from a previous declaration is used for other generic entity
types that have redeclarations.
When a function declaration is qualified, such as defining a class
method outside the class body, we need only the function declaration to
contribute to its generic region stack. The code was collecting constant
values from all qualifier segments together incorrectly.
So when we PushNameQualifierScope(), we also drop the current generic
region stack and rewrite its constant values by calling
FinishGenericRedecl(), and open a new stack entry for the next part of
the qualified declaration.
If a generic declaration somehow has more dependent instruction than a
previous declaration, it would add new instructions to its eval block
with indices beyond the elements in the actual declaration eval block,
since we only store the block from the first declaration found. To avoid
this we plumb through that we are in a redeclaration, and terminate with
an ICE instead of adding new instructions to crash on later.
Fixes#5136.
The type structure is built for the impl lookup query for the
combination of the self type and the interface being queried. Then it is
built for each `impl` definition that is a potential candidate.
The type structures are compared to ensure they have a compatible
structure, and the `impl` declaration is not considered if they do not.
Finally, the type structures are used as a sorting key for the candidate
`impl` declarations, with the most-specified type structures (the ones
with the furthest distance to the first symbolic value) coming first in
the ordering.
See
https://docs.carbon-lang.dev/docs/design/generics/overview.html#parameterized-impl-declarations
for the design of the type structure and related ordering.
Most of the commits in this PR landed in #5158 (11ae0e27ab) by
mistake, but this includes the final changes since that PR was written.
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
When the parameter is a deduced symbolic FacetValue, refering to a
BindSymbolicName, and the argument is a concrete FacetValue that would
match the FacetType requirements on the BindSymbolicName's type, we
currently do not deduce that the argument matches the parameter.
The argument is not _converted_ to the parameter type because they are
both FacetValues of the same FacetType type. However they are also not
equal constant values so the argument is not saved as a deduced match
for the parameter.
In order to accept the FacetValue, we need to consider them as
`deduce_through`, which attempts to deduce each of the fields in the
argument FacetValue against the fields in the parameter FacetValue.
This deduces that the argument's concrete type matches the symbolic
BindSymbolicName and its witnesses are the same.
Since the parameter is a FacetValue, its argument is not the type that
needs to be recorded as the deduced type for the binding. The
BindSymbolicName inside the parameter is the place that we need to find
the deduced type for the binding. So simply walking into the FacetValue
gets us to that position, where we eventually record the deduced
argument type as being the concrete type from the original argument
FacetValue.
Similarly, when determining what interfaces are satisfied by a
FacetValue for deduce, we want to use the full type available in the
FacetValue rather than just those from its FacetType. Determining
availability of interfaces here is equivalent to converting, and we want
converting a FacetValue to always work on the full available type info.
Only API access (member lookup) is restricted by a FacetValue to the
interfaces provided by its FacetType type.
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
This changes the SemIR of invalid redeclarations, because previously
they lacked a name. We've avoided this in diagnostics so it doesn't
otherwise come up, but I plan to use it for more easily validating
redeclarations.
A query facet type may contain multiple required interfaces, in which
case impl lookup should return an ImplWitness for an impl that is used
for each interface in the query. We bundle these together into an
instruction block and return that from impl lookup. The witnesses are in
the same order as the interfaces in the
`CompleteFacetType::required_interfaces`. This allows walking the
`required_interfaces` to find an interface to give an index that can
also be used to grab a witness from this set, or from FacetValue.
FacetValue now has an InstBlockId for the set of witnesses of the
FacetType, instead of a single ImplWitness instruction id.
FacetAccessWitness includes the index of the witness (determined from
the position in `required_interfaces`) of the witness it's accessing
from the FacetType.
The
toolchain/check/testdata/facet/no_prelude/fail_todo_call_combined_impl_witness.carbon
test demonstrates the fix in the resulting SemIR. We can see the calls
to methods on a multi-interface FacetType result in a FacetAccessWitness
with an index of the correct interface, and this results in a witness
that leads to the correct impl's function.
There is a TODO in member access, where it does not have a
`CompleteFacetType` yet, so it uses the index in
`FacetTypeInfo::impls_constraints` instead, but this can be incorrect in
the presence of named constraints, which when completed can add more
interfaces to the `CompleteFacetType` and which are sorted into an
arbitrary order with the rest there.
---------
Co-authored-by: josh11b <15258583+josh11b@users.noreply.github.com>
During SemIR, when identifying a specific in a specific context, we'd
have to either look through a specific or through a bound method.
Canonicalize which one to look through first, by having the BoundMethod
created around a SpecificFunction instead.
This changes a lot of check tests.
TODO: As the SemIr does not currently allow removal (access to insts()
is intentionally const), the bound instruction created prior to finding
the specific is not removed from the instructions.
Options: (1) leave as is, (2) add a way to remove the previous bound,
(3) rethink how/when the BoundMethod inst is created.