This switches `DCHECK` and `FATAL` as well.
The goal is to reduce the code size impact of these assertions so that
we can keep more of them enabled. Currently, the largest cost I see from
`CHECK` is not the actual check or the cold code itself, but actually
the failure to inline trivial functions due to the presence of the cold
code. This means that our goal isn't to reduce apparent code size in the
final binary but the LLVM IR cost assessed for these routines in the
inliner, which closely correlates with code size but is a bit different.
As discussed in #4283, experimentation shows that a single function call
with a minimal number of arguments is the lowest cost model for these.
This is easily achieved with a format-string API that internally uses
`llvm::formatv`. This PR is essentially the `CHECK` version of #4283.
However, the check macros are substantially harder to make work with
both format strings and streaming because they also take a condition.
Also, unexpectedly, I was very successful at devising a regular
expression based automated rewrite from the streaming to the format
string form with only low 10s of manual fixes. This includes compacting
strings broken up across lines, etc. Given how well that went, I've
prepared this PR which just directly switches to the format string API
and migrate everything to use it.
One nice side-effect is that the format string approach ends up greatly
simplifying the implementation here as well.
This is ... *shockingly* effective. Parsing speeds up by more than 3%
with just this change. And checking speeds up by **8%** with this change
alone:
```
BM_CompileAPIFileDenseDecls<Phase::Parse>/256 86.3µs ± 1% 82.9µs ± 1% -3.94% (p=0.000 n=17+19)
BM_CompileAPIFileDenseDecls<Phase::Parse>/1024 431µs ± 1% 415µs ± 1% -3.76% (p=0.000 n=18+19)
BM_CompileAPIFileDenseDecls<Phase::Parse>/4096 1.77ms ± 1% 1.71ms ± 1% -3.18% (p=0.000 n=18+19)
BM_CompileAPIFileDenseDecls<Phase::Parse>/16384 7.44ms ± 1% 7.17ms ± 2% -3.56% (p=0.000 n=18+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/65536 30.7ms ± 1% 29.7ms ± 1% -3.15% (p=0.000 n=18+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/262144 131ms ± 1% 127ms ± 1% -2.81% (p=0.000 n=18+18)
BM_CompileAPIFileDenseDecls<Phase::Check>/256 878µs ± 2% 800µs ± 1% -8.91% (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/1024 1.88ms ± 2% 1.72ms ± 1% -8.56% (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/4096 5.78ms ± 2% 5.28ms ± 1% -8.70% (p=0.000 n=20+18)
BM_CompileAPIFileDenseDecls<Phase::Check>/16384 21.9ms ± 1% 20.1ms ± 1% -8.02% (p=0.000 n=18+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/65536 90.4ms ± 2% 83.1ms ± 1% -8.04% (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/262144 381ms ± 2% 352ms ± 1% -7.79% (p=0.000 n=19+19)
```
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Co-authored-by: josh11b <15258583+josh11b@users.noreply.github.com>
With the toolchain splitting namespaces, ostream.h's `operator<<`
templates aren't reliably found with name lookup, likely due to the loss
of associated namespaces (zygoloid commented on this at
https://github.com/carbon-language/carbon-lang/pull/3161#discussion_r1307941999).
This is especially a barrier to moving the lex files into `Carbon::Lex`;
versus other parts of the toolchain, they contain more printable types
which are used cross-namespace, including `Carbon::Testing`. As a
consequence, I'm looking at migrating ostream.h to a more reliable
approach that doesn't rely as much on everything being in the `Carbon`
namespace.
See arena.h for discussion of what canonicalization means in this
context. This is primarily intended to support implementing a memo table
of template instantiations to resolve#2951, but could be useful for
other purposes as well.
Additional changes:
- Pass `VTable` constructor parameters by pointer, to avoid the need to
define `operator==` and `hash_value` for it.
- Clean up the recurring pattern of allocating identical `NamedElement`s
on the stack and heap. Instead we always allocate it on the heap and
pass it by pointer.
- Add `Print()` and `Dump()` to `Bindings` as a debugging convenience.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Prevent copies when initializing value expression from reference
expression. This is based on
https://github.com/carbon-language/carbon-lang/pull/2006, which
introduces expression categories, and how it is possible to convert
to/from those different categories. Continuation of
https://github.com/carbon-language/carbon-lang/pull/2907
## Functional changes
* Initializing a value expression from a reference expression takes its
value without a copy
* Reading from the value expression causes an error if the value changed
from the time it was initialized
* In this situation, prevents a copy both for variable definitions, and
call parameter bindings
## Main implementation changes
* Add new `ExpressionCategoryAction`, which evaluates an expression and
returns an `ExpressionValue` containing its category and address (if
any), in addition to the resulting `Value*`
* `ExpressionAction`s now invokes `ExpressionCategoryAction` and unwraps
the returned `ExpressionValue`
* `RuntimeScope::BindAndPin` method, and corresponding when attempting
to read a `value_node`.
## Next work
* Avoid unnecessary copies from value expression to value expression,
after ensuring that even value expression temporaries are registered for
destruction (https://github.com/Pixep/carbon-lang/pull/9)
### Features
* Add virtual `destructor`s support (Closes#2521)
* Check virtual override for virtual destructors
* Error if attempting to `Delete` a class that does not have virtual destructors from a base class pointer
### Implementation
* Update parser to support virtual override introducers for destructors
* Check virtual override for class destructor and add to class vtable if necessary
* Add corresponding tests
### Notes
Contrary to initial implementation, this implementation leverages the `Address` structure and implements a new `Address::DowncastedAddress()` method to get address from child most class from a base class address. This avoids the need to use `GetAllocationId` and its issues when it comes to having multiple values for an `AllocationId`.
### Next work
Following this PR, we need to:
* Check when using `Delete` that the class was allocated with `New` (WIP)
* Drop the old `GetAllocationId(Value*)` in favor of a better system (WIP)
These are used by the AST in lots of ways, and this resolves various layering issues.
This means that `AllocationId` also lives in ast/, but is managed by interpreter/. A better layering here would be desirable, but this seems good enough for the time being.
Jon Ross-Perkins
Adrien Leravat