This change adds another file for benchmarking luau-analyze and sets up
benchmarks for both non-strict/strict modes for analysis and all three
optimization levels for compilation performance.
To avoid issues with race conditions on repository update we do all this
in the same job in benchmark.yml.
To be able to benchmark both modes from a single file, luau-analyze
gains --mode argument which allows to override the default typechecking
mode. Not sure if we'll want this to be a hard override on top of the
module-specified mode in the future, but this works for now.
Since callgrind allows to control stats collection from the guest, this
allows us to reset the collection right before the benchmark starts.
This change exposes this to the benchmark runner and integrates
callgrind data parsing into bench.py, so that we can run bench.py with
--callgrind argument and, as long as the runner was built with callgrind
support, we get instruction counts from the run.
We convert instruction counts to seconds using 10G instructions/second
rate; there's no correct way to do this without simulating the full CPU
pipeline but it results in time units on a similar scale to real runs.
* Fix free Luau type being fully overwritten by 'any' and causing UAF
* Fix lua_clonefunction implementation replacing top instead of pushing
* Falsey values other than false can now narrow refinements
* Fix lua_getmetatable, lua_getfenv not waking thread up
* FIx a case where lua_objlen could push a new string without thread wakeup or GC
* Moved Luau math and bit32 definitions to definition file
* Improve Luau parse recovery of incorrect return type token
This meta-RFC proposes removing the previously accepted RFC on safe navigation operator. This is probably going to be disappointing but is the best course of action that reflects our ideals in language evolution.
See PR thread for rationale and discussion.
* Run clang-format
* Contains a preliminary implementation of deferred constraint resolution
* Reduce stack usage by some recursive functions
* Fix a bug when smartCloning a BoundTypeVar
* Remove some GC related flags from VM
* Recap is ready
* Change the intro format so it shows up in the news section snippet
* Fixed line breaks that were implied
* Update docs/_posts/2022-06-01-luau-recap-may-2022.md
Co-authored-by: Arseny Kapoulkine <arseny.kapoulkine@gmail.com>
* Drop Autocomplete improvements and extend compiler optimization section
In the cross-post to Roblox developer forum, Autocomplete section will be restored, while compiler optimization section can be optionally removed (although it might interest some developers)
* One more optimization
* Update docs/_posts/2022-06-01-luau-recap-may-2022.md
Co-authored-by: dcope-rbx <91100513+dcope-rbx@users.noreply.github.com>
Co-authored-by: Arseny Kapoulkine <arseny.kapoulkine@gmail.com>
Co-authored-by: dcope-rbx <91100513+dcope-rbx@users.noreply.github.com>
* Adds a currently unused x86-64 assembler as a prerequisite for possible future JIT compilation
* Fix a bug in table iteration (closes Possible table iteration bug #504)
* Improved warning method when function is used as a type
* Fix a bug with unsandboxed iteration with pairs()
* Type of coroutine.status() is now a union of value types
* Bytecode output for tests/debugging now has labels
* Improvements to loop unrolling cost estimation
* Report errors when the key obviously doesn't exist in the table
* Respect useLineBreaks for union/intersect toString
* Apply suggestions from code review
Co-authored-by: Andy Friesen <andy.friesen@gmail.com>
Co-authored-by: Andy Friesen <andy.friesen@gmail.com>
Adds an option to hide the `self: type` argument as the first argument in the string representation of a named function type var if the ftv hasSelf.
Also added in a test for the original output (i.e., if the option was disabled)
I didn't apply this option in the normal `Luau::toString()` function, just the `Luau::toStringNamedFunction()` one (for my usecase, that is enough + I felt like a named function would include the method colon `:` to signify self). If this is unintuitive, I can also add it to the general `Luau::toString()` function.
Update `__pairs` note with `__iter`, change `__len` to unsure as with `__iter` lack of `__len` on tables is the only issue preventing complete user created containers.
Update the typecheck.md page to talk about singleton types and their uses, tagged unions.
As a driveby, improve the documentation on type refinements. And delete the unknown symbols part, this is really dated.
* Update docs/_pages/typecheck.md to fix a typo
Co-authored-by: Arseny Kapoulkine <arseny.kapoulkine@gmail.com>
This is a meta-RFC. I'd like to propose to remove the postfix `!` operator from the language.
I'd like to argue that this operator will eventually become useless and can only be used incorrectly at some point in the future, and that there are likely ways for them to be sound without having to reach for the `!` hammer.
With that, I present these counterarguments:
Shortcoming <span>#</span>1: Refinements does not apply to the block after `return`/`break`/`continue`.
```lua
if not x or not y then return end
-- both x and y are truthy
```
This will be solved by implementing control flow analysis where it'd apply the inverse of the condition leading up to the control transfer statement to the rest of the scope.
Shortcoming <span>#</span>2: Type checker is not aware of the actual state of various locations.
```lua
type Foo = { x: { y: number }? }?
local foo: Foo = { x = { y = 5 } }
print(foo.x.y) -- prints 5 at runtime, type checker warns on this
```
This will be solved by implementing type states where it would inspect the initialization sites as well as assignments to know their actual states. That is, rather than trusting the type annotation `Foo` as the state which gets us far enough, we'd start seeing these type annotations as a subtype constraint for the location `foo`.
---
If there are other use cases not covered in this message, we should talk about that and see if there exists an alternative direction that can solve these use cases soundly.