M-Labs
/
nac3-spec
11
0
Fork 0
Code Issues 14 Pull Requests Releases Wiki Activity

Updated specification #8

Merged
pca006132 merged 5 commits from updated into master 2021-06-07 10:06:52 +08:00
Conversation 5 Commits 5 Files Changed 1 +41 -85
1 changed files with 41 additions and 85 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit d82edd2753 - Show all commits

126
README.md
View File

@ -5,23 +5,22 @@ Specification and discussions about language design.
A toy implementation is in [`toy-impl`](./toy-impl), requires python 3.9. A toy implementation is in [`toy-impl`](./toy-impl), requires python 3.9.
## Referencing Python Variables ## Referencing Host Variables from Kernel
The kernel is allowed to read host Python variables, but has to specify with Host variable to be accessed must be declared as `global` in the kernel
pca006132 marked this conversation as resolved
lriesebos commented 2021-04-12 04:19:36 +08:00
Review
Copy Link

I am still confused what the idea is about the use of 'global'. Is this what you mean?

FOO = 0

@kernel
def correct() -> int:
    global FOO
    return FOO + 1
    
@kernel
def fail_without_global() -> int:
    return FOO + 2
    
@kernel
def fail_write() -> None:
    FOO += 1
I am still confused what the idea is about the use of 'global'. Is this what you mean? ```python FOO = 0 @kernel def correct() -> int: global FOO return FOO + 1 @kernel def fail_without_global() -> int: return FOO + 2 @kernel def fail_write() -> None: FOO += 1 ```
pca006132 commented 2021-04-12 15:24:53 +08:00
Review
Copy Link

yes

yes
`global` before referencing them. This is to simplify and speed-up function. This is to simplify and speed-up implementation, and also warn the
implementation, and also warn the user about the variable being global. (prevent user about the variable being global. (prevent calling the interpreter many
calling the interpreter many times during compilation if there are many times during compilation if there are many references to host variables)
references to host variables)
Only primitive types and tuple/list of primitive types are allowed. Kernel cannot modify host variables, this would be checked by the compiler.
Value that can be observed by the kernel would be frozen once the kernel has
been compiled, subsequence modification within the host would not affect the
sb10q commented 2021-04-12 16:06:03 +08:00
Review
Copy Link

Ah, this would cause another problem like those in #5
But deleting globals or trying to make them immutable in the interpreter sounds messy, and a global being modified in a RPC seems to be a niche case.
So I'm ok with this current proposal, the behavior simply should be documented.

Ah, this would cause another problem like those in https://git.m-labs.hk/M-Labs/nac3-spec/issues/5 But deleting globals or trying to make them immutable in the interpreter sounds messy, and a global being modified in a RPC seems to be a niche case. So I'm ok with this current proposal, the behavior simply should be documented.
pca006132 commented 2021-04-12 18:05:01 +08:00
Review
Copy Link

Yes, I was thinking about this either, but it seems like there is no easy way around this.

Yes, I was thinking about this either, but it seems like there is no easy way around this.
lriesebos commented 2021-04-12 23:42:58 +08:00
Review
Copy Link

Why don't we force users to type globals used in kernels as Invariant[]? The semantics are the same, making it clear to the user how this global variable can be used in kernels.

Why don't we force users to type globals used in kernels as `Invariant[]`? The semantics are the same, making it clear to the user how this global variable can be used in kernels.
sb10q commented 2021-04-13 00:01:08 +08:00
Review
Copy Link

It cannot be enforced in the interpreter.

It cannot be enforced in the interpreter.
lriesebos commented 2021-04-13 00:14:08 +08:00
Review
Copy Link

I understand, it would just potentially make it more clear to the user that the global variable is immutable in the kernel. But I am fine with either!

I understand, it would just potentially make it more clear to the user that the global variable is immutable in the kernel. But I am fine with either!
pca006132 commented 2021-04-13 10:04:01 +08:00
Review
Copy Link

I think the types that we allow are basically immutable? Int, tuples, string, etc.
Even if they are decorated with Invariant[T], the user can just re-assign them, so hard to enforce.

I think the types that we allow are basically immutable? Int, tuples, string, etc. Even if they are decorated with `Invariant[T]`, the user can just re-assign them, so hard to enforce.
sb10q commented 2021-04-13 10:16:53 +08:00
Review
Copy Link

For classes there are special __getattr__ and __setattr__ methods that we can use to hack the interpreter and enforce KernelImmutable restrictions.
There is no equivalent for modules/globals.

For classes there are special ``__getattr__`` and ``__setattr__`` methods that we can use to hack the interpreter and enforce ``KernelImmutable`` restrictions. There is no equivalent for modules/globals.
lriesebos commented 2021-04-13 10:26:56 +08:00
Review
Copy Link

Yes, @sb10q showed a POC for immutable instance variables in #5, and I agree this does not work for global variables.

For me, the reasoning was to type global variables as KernelImmutable[] because they are per definition kernel immutable. That would align their type annotation with instance variables. Though since we can not enforce that at runtime, globals will anyway be a special case that needs documentation. So as mentioned before, I am also fine with not typing it KernelImmutable[].

Yes, @sb10q showed a POC for immutable instance variables in #5, and I agree this does not work for global variables. For me, the reasoning was to type global variables as `KernelImmutable[]` because they are per definition kernel immutable. That would align their type annotation with instance variables. Though since we can not enforce that at runtime, globals will anyway be a special case that needs documentation. So as mentioned before, I am also fine with not typing it `KernelImmutable[]`.
kernel.
The value would be substituted at compile time, subsequent modification in the Only types supported in the kernel can be referenced.
host would not be known by the kernel.
Modification in kernel code to the global variables is not allowed.
## Class and Functions ## Class and Functions
* Class fields must be annotated: * Instance variables must be annotated: (Issue #1)
```py ```python
class Foo: class Foo:
a: int a: int
b: int b: int
@ -29,24 +28,24 @@ Modification in kernel code to the global variables is not allowed.
self.a = a self.a = a
self.b = b self.b = b
``` ```
* Three types of instance variables: (Issue #5)
* Host only variables: Do not add type annotation for it in the class.
* Kernel Invariants: Immutable in the kernel and in the host while the kernel
is executing. Type: `KernelInvariant(T)`. The types must be immutable.
(use tuple instead of list in the host, but the type annotation should still
be list?)
* Normal Variables: The host can only assign to them in the `__init__`
function. Not accessible afterwards.
* Functions require full type signature, including type annotation to every * Functions require full type signature, including type annotation to every
parameter and return type. parameter and return type.
```py ```python
def add(a: int, b: int) -> int: def add(a: int, b: int) -> int:
return a + b return a + b
``` ```
* No implicit coercion, require implicit cast. Integers are int32 by default,
floating point numbers are double by default.
* RPCs: optional parameter type signature, require return type signature. * RPCs: optional parameter type signature, require return type signature.
* Function default parameters are not allowed, as changes to the default value * Function default parameters must be immutable.
would not be kept across kernel calls, which is a potential source of * Function pointers are supported, and lambda expression is not supported
confusion. (maybe we can allow primitive default types?) currently. (maybe support lambda after implementing type inference?)
* Cannot construct objects within kernel code.
Questions:
* Can we construct objects within kernel code?
* Should we support function pointers? What about subtyping with function
pointers, and generic types?
## Built-in Types ## Built-in Types
* Primitive types include: * Primitive types include:
@ -57,14 +56,14 @@ Questions:
* `uint32` * `uint32`
* `uint64` * `uint64`
* `float` * `float`
* `str` (note: fixed length, provide builtin methods?) * `str`
* `bytes` (a list of `byte`, but with more convenient syntax) * `bytes`
* Collections include: * Collections include:
* `list`: homogeneous (elements must be of the same type) fixed-size (no * `list`: homogeneous (elements must be of the same type) fixed-size (no
append) list. append) list.
* `tuple`: inhomogeneous fixed-size list, only pattern * `tuple`: inhomogeneous immutable list, only pattern
matching (e.g. `a, b, c = (1, True, 1.2)`) and constant indexing matching (e.g. `a, b, c = (1, True, 1.2)`) and constant indexing is
is supported: supported:
``` ```
t = (1, True) t = (1, True)
# OK # OK
@ -75,23 +74,25 @@ Questions:
i = 0 i = 0
a = t[i] a = t[i]
``` ```
* `range` (over numerical types) (not sure if this is really useful) * `range` (over numerical types)
### Numerical Types ### Numerical Types
* All binary operations expect the values to have the same type, no implicit * All binary operations expect the values to have the same type.
coercion would be performed, explicit casting is required.
* Casting can be done by `T(v)` where `T` is the target type, and `v` is the * Casting can be done by `T(v)` where `T` is the target type, and `v` is the
original value. Examples: `int64(123)` original value. Examples: `int64(123)`
* Constant integers are treated as `int32` by default. If the value cannot * Integers are treated as `int32` by default. Floating point numbers are double
be stored in `int32`, `uint64` would be used if the integer is non-negative, by default.
and `int64` would be used it the integer is negative. * No implicit coercion, require implicit cast.
For integers that don't fit in int32, users should cast them to `int64`
explicitly, i.e. `int64(2147483648)`. If the compiler found that the integer
does not fit into int32, it would raise an error. (Issue #2)
* Only `uint32`, `int32` (and range of them) can be used as index. * Only `uint32`, `int32` (and range of them) can be used as index.
## Generics ## Generics
We use [type variable](https://docs.python.org/3/library/typing.html#typing.TypeVar) for denoting generics. We use [type variable](https://docs.python.org/3/library/typing.html#typing.TypeVar) for denoting generics.
Example: Example:
```py ```python
from typing import TypeVar from typing import TypeVar
T = TypeVar('T') T = TypeVar('T')
@ -102,16 +103,13 @@ class Foo(EnvExperiment):
return a == b return a == b
``` ```
* Type variables can only be used in functions/methods, but not in classes.
(this can be relaxed, only allow those with type variables fully defined
from the constructor)
* Type variable can be limited to a fixed set of types. * Type variable can be limited to a fixed set of types.
* Type variables are invariant, same as the default in Python. We disallow * Type variables are invariant, same as the default in Python. We disallow
covariant or contravariant. The compiler should mark as error if it encounters covariant or contravariant. The compiler should mark as error if it encounters
a type variable used in kernel that is declared covariant or contravariant. a type variable used in kernel that is declared covariant or contravariant.
* Code region protected by a type check, such as `if type(x) == int:`, would * Code region protected by a type check, such as `if type(x) == int:`, would
treat `x` as `int`, similar to how [typescript type guard](https://www.typescripttutorial.net/typescript-tutorial/typescript-type-guards/) works. treat `x` as `int`, similar to how [typescript type guard](https://www.typescripttutorial.net/typescript-tutorial/typescript-type-guards/) works.
```py ```python
def add1(x: Union[int, bool]) -> int: def add1(x: Union[int, bool]) -> int:
if type(x) == int: if type(x) == int:
# x is int # x is int
@ -131,48 +129,6 @@ Questions:
* Would it be better to assert on the type variable directly instead of * Would it be better to assert on the type variable directly instead of
`type(x)` for type guards? `type(x)` for type guards?
## Dynamic Dispatch ## Lifetime
Type annotations are invariant, so subtype (derived types) cannot be used Probably need more discussions...
when the base type is expected. Example:
```py
class Base:
def foo(self) -> int:
return 1
class Derived(Base):
def foo(self) -> int:
return 2
def bar(x: list[Base]) -> int:
sum = 0
for v in x:
sum += v.foo()
return sum
# incorrect, this list cannot be typed (inhomogeneous)
bar([Base(), Derived()])
```
Dynamic dispatch is supported, but requires explicit annotation, similar to
[trait object](https://doc.rust-lang.org/book/ch17-02-trait-objects.html) in rust.
`virtual[T]` is the type for `T` and its subtypes(derived types).
This is mainly for performance consideration, as virtual method table that is
required for dynamic dispatch would penalize performance, and prohibits function
inlining etc.
Type variables cannot be used inside `virtual[...]`, and type variables would not
range over `virtual[...]`.
> Not sure what is the best syntax for `virtual[...]`
Example:
```py
def bar2(x: list[virtual[Base]]) -> int:
sum = 0
for v in x:
sum += v.foo()
return sum
```