When deducing arguments for generic parameters of an `impl`, the
deduction calls `Convert` on the input arguments. Often, the input
argument is a facet, and needs to be converted to a type via
FacetAccessType in order to produce a different facet. These
instructions end up being added to the semir, but only their constant
values are needed for the resulting specific returned from Deduce.
In the best case, these extra instructions are just noise in the semir,
or they just cause instruction names to get differentiated with larger
suffixes.
In the worst case, these extra instructions contain references to
instructions from a generic context, and leak them out of that generic
context and into another. In particular, when importing a
LookupImplWitness instruction, the re-evaluation of it can do deduce
(when the lookup is against a generic `impl`). The instructions created
in Deduce are not part of the import, and end up referring to imported
instructions from the local context, which leads to confusion in the
toolchain, and can crash.
The `import_self_specific.carbon` test demonstrates this. It causes the
`I.F` function to be imported from the `I` interface when building the
witness table for the `impl`. Doing so imports the specific of `C` which
includes a LookupImplWitness for `Self.Accoc` in `I`. The `Self` is a
BindSymbolicName with generic binding index 0, in `I`. When Convert
creates instructions in the generic `impl forall D`, however, they end
up referencing and including this BindSymbolicName into its eval block.
But the generic binding 0 in the `impl` is a very different thing (a
value of type `E`). This confusion leads to crashes.
The SymbolicBindingType refers to the type value that will be
substituted in for the BindSymbolicName, but holds onto the EntityNameId
from the BindSymbolicName instead of (or in addition to, for now) the
instruction.
The EntityNameId will be used to look in the ScopeStack to find the
witnesses either from the BindSymbolicName instruction, or other
instructions that specify `impls` constraints against the EntityName.
This will allow us to have the `T` in `I(T)` resolve to a `.Self`
reference in the type so that we get type equality with the binding's
type: `T:! I(.Self)`.
This also does a little restructuring in the same direction, following
#6124.
Leads want `Destroy` to work similarly now for all types. As a
consequence, there doesn't seem to be as much benefit to splitting off
aggregate destruction. In this PR, the `type.destroy` function can now
be expected to destroy anything that's destructible; that means it'll be
usable for the `final fn` once that support is available.
Similarly, this gets rid of the impls other than the single blanket
impl, now using `type.can_destroy`. Since they all need to use the same
function, there's no benefit to splitting approaches. Also, now it can
just be a `final impl` since there should be no need for people to
create specializations -- if this blanket impl applies, it means the
`final fn` is the same.
This also slips in `partial` support since there's no reason to have it
diverge anymore. Also `abstract`, which I'm not sure is broadly testable
since most cases it'd come up, the `abstract` keyword is explicitly
detected/rejected.
Note though that this doesn't make any really big changes. It's just
realigning on the leads decision. I'm going this way to try to reduce
name-related churn for other changes.
The main direction of this change is the edits to `destroy.carbon`
(matching in both prelude and min_prelude).
Previously there was a no-op blanket impl for `Destroy`, which hid all
missing implementations of `Destroy`. This does a few things:
- Sets up builtin aggregate destruction for struct and tuple types as
before, but also adds C++ class types and array types to the same
handling. (all as a TODO for actual implementation)
- Also maybe-unformed destruction, for now at least. (there's a chance I
may try a different approach on this, but the impl lookup wasn't working
as I'd hope in order to write it in code)
- Adds handlers for simple things that are easy to do in code: `type`,
`bool`, pointers. (because these are no-op destruction)
- Redirect `const T` destruction to `T` destruction.
This leaves as future issues:
- `partial T` destruction. (this can't be done similar to `const`
because it only works for non-`final` class types; I think `class`
definitions should just generate what's needed)
- Destruction of other prelude-provided types. (will probably come up as
we implement class destruction, that the adapted builtin type doesn't
implement `Destroy` -- but may end up special-casing that in a way that
moots it)
This moves the `&` operator from `facet_types.carbon` to
`convert.carbon` because more things need to handle type and now that
we're getting separate copy and destroy interfaces. It should be
low-cost (an interface and builtin) so hopefully this is the right
balance for complexity and re-use.
A few tests are also edited in order to focus them more on what they
intend to test, and avoid a `Destroy` dependency.
This is in support of a goal of changing the blanket `destroy` impl to
use (roughly):
```
private fn CanAggregateDestroy() -> type = "type.can_aggregate_destroy";
// Handles aggregate type destruction.
impl forall [AggregateDestroyT:! CanAggregateDestroy()] AggregateDestroyT as Destroy {
fn Op[addr self: Self*]() = "type.aggregate_destroy";
}
```
That isn't done here because there's still other issues that migrating
raises. What this *does* do is add the builtin functions, and in
particular, support to `FacetTypeInfo` to make `CanAggregateDestroy`
work.
The "special requirement" approach in `FacetTypeInfo` allows us to
support restricting a blanket impl under the current approach of impls.
Maybe we'll find a cleaner approach that can work in the future, but
this fits into the current model by propagating similar to other
requirements. I'm using an enum mask because we have a number of similar
things to add (e.g. copy, move) but I'm not sure we need a full vector.
A few alternatives considered were:
- Supporting syntax more like `where .Self impls
TypeCanAggregateDestroy(.Self, SupportedInterface,
UnsupportedInterface)`. I think it'd be a little cleaner, but requires
better compile-time evaluation in order to assess the type of the call.
Right now it's expected to be a `FacetType` too early to make this work,
and I was concerned about pouring too much more time down this route.
- Providing an actual interface, in particular doing name lookup back
into `Core.` for an interface. This would've added name lookup overhead,
and the question of whether an `impl` exists.
- Generating an interface. This avoids the name lookup, but would still
raise the question of whether an `impl` should also be generated. Work
I've previously done generating interfaces for class destruction also
feels complex to both write and understand (an unfortunate issue).
- Still modeling as an `ImplsConstraint`, for example by defining a
special `InterfaceId::CanAggregateDestroy = -2` similar to what we do on
other ids. I was hesitant because of how this expands the number of
modes of `InterfaceId`, and things for consuming code to watch out for,
for what feels like a relatively niche set of use-cases that are only
interface-like.
---------
Co-authored-by: Dana Jansens <danakj@orodu.net>
When returning a value from a function whose return type has a by-copy
initializing representation, perform initialization like we do when the
return type has an in-place initializing representation. This makes our
SemIR representation more uniform, as the return expression will now
always be an initializing expression rather than a value expression, but
more importantly it means that attempts to return a non-copyable type by
value now fail, even if the type has a by-copy initializing
representation.
This catches a bunch of places where we were returning a value of an
unconstrained template parameter `T:! type`, which we were incorrectly
allowing because we didn't notice it was not copyable. Unfortunately
this then requires quite a few test updates.
Like #6034, this exposes a lowering issue where lowering crashes when
attempting to lower a specific copy operation for certain types; a
couple more tests are temporarily disabled here. An upcoming PR
dependent on this one will fix the issue and re-enable those tests.
This uses each vector's size as a barrier between lists, to eliminate
the possibility of incidental collisions between entries of different
lists. This is the same as is done inside `AddBlock`.
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>
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 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>
Now that included files can specify `EXTRA-ARGS`, this uses that to
handle min_prelude files. Also moves `As`/`ImplicitAs` out to a shared
file, partly because we duplicate it a few times over, partly just to
show that it works.
Also removes the `min_prelude/` subdirectories because many of these
files were touched by autoupdate regardless. I noticed one conflict for
`impl_thunk.carbon`, so renaming that one to
`impl_thunk_min_prelude.carbon`.
`GetArgReplacements` I simply noticed was unused, so removing it.
This could probably go further, but I'm just trying to get the ability
to run some significant tests without the prelude. This already requires
some migration of `i32` and `bool` to non-prelude-dependent types
(technically not at the moment, but I'm trying to stick with the
philosophical model of the prelude).
Note this undoes part of #3895 which had made all lex/parse tests depend
on the prelude -- I don't think that was a necessary change. Rather, it
seems better to isolate individual lex/parse tests from the prelude. I'd
been on the fence in that PR (it was fixing one of the issues that
contributed to wanting to not include the prelude), but now it feels
more consistent. The prelude could have a parse error and we shouldn't
break every parse test on that.
Factor out `SemIR::InstNamer` and also use it when lowering to LLVM IR.
Automatically name all instructions created with our `IRBuilder` based
on the name computed by the `InstNamer`, and likewise name basic blocks
using the label generated by the `InstNamer`.
Move some of the existing naming logic out from lower into `InstNamer`
so that it's also used in SemIR. In particular, we now name call
instructions after their callee, or after the builtin name for calls to
builtins.
Computing and adding these names isn't completely free. This instruction
naming is designed to be optional, so that we can turn it off for builds
where the LLVM IR will only be converted to assembly and won't be seen
by a human, but so far it's enabled unconditionally. We can tune that
later as needed.
Move completeness check to the point where the function is defined or
first called. This means we also defer deciding whether the function has
a return slot until that point. Instead of storing a return slot per
function, store the location of the return storage, which may or may not
be used, and compute and store a separate flag saying whether to use it
at the point of first use or definition.
This is the final piece in supporting simple `Make` functions in classes
as a replacement for constructors.
- Adds an empty prelude.carbon file
- Imports that file in any non-Core package file
- Adds --disable-prelude-import to avoid that
- Adds --exclude-dump-file-prefix to be able to hide files from dumping
- Used to hide core files (we can't do this by package name due to lex
dumps, for example)
- Restructures some tests to not rely on `i32`, particularly `alias`
tests (which rely on a name ref) and tests with no prelude.
I'm adding the framework for switching i32 to calling Int32 in the
prelude, but I'm running into a separate error actually switching over.
So that *mostly* works, but isn't quite ready for prime time. However,
maybe the current state of this PR is still useful to review since it
does a lot of the infrastructure work and adds the %Core everywhere?
Previously we used the default inst block formatting, which writes out a
parenthesized list of references, which would always be 'unexpected
instref's because nothing else prints the instructions in the decl
block.
In addition, track the decl block for function declarations like we do
for other kinds of declaration. This means that the parameter
declarations for a function are now properly rendered into the formatted
IR. Note that this adds a lot of verbosity to `function_decl`, but it
does accurately reflect the IR, and we'll probably want this information
to be printed once we start supporting more complex generic function
declarations.
This removes almost all the 'unexpected instref's in our formatted
output. There are remaining cases when a declarative scope contains
multiple blocks, where we only track one of those blocks. That happens
when there is control flow within declarative scopes, and for error
recovery when a class or interface or similar is defined more than once.
As requested in #3730.
Adds `BindAlias` with a hybrid of `BindName` and `NameRef` semantics. I
think it's slightly closer to `BindName` because it introduces a name,
so I'm going more in that direction. This also matches the need for
`bind_name_id` with imports on enclosing scopes.
Note, only things that look like a name reference are being allowed on
the RHS of `alias`. This includes builtins that look like name
references, such as `bool`, but not ones that turn into values
underneath, such as `false`.
Dana Jansens
Jon Ross-Perkins
Richard Smith
