lighthouse/consensus/ssz_derive/src/lib.rs

623 lines
21 KiB
Rust
Raw Normal View History

#![recursion_limit = "256"]
//! Provides procedural derive macros for the `Encode` and `Decode` traits of the `eth2_ssz` crate.
//!
//! Supports field attributes, see each derive macro for more information.
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
use darling::{FromDeriveInput, FromMeta};
use proc_macro::TokenStream;
2019-04-17 11:21:07 +00:00
use quote::quote;
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
use std::convert::TryInto;
use syn::{parse_macro_input, DataEnum, DataStruct, DeriveInput, Ident};
/// The highest possible union selector value (higher values are reserved for backwards compatible
/// extensions).
const MAX_UNION_SELECTOR: u8 = 127;
#[derive(Debug, FromDeriveInput)]
#[darling(attributes(ssz))]
struct StructOpts {
#[darling(default)]
enum_behaviour: Option<String>,
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
/// Field-level configuration.
#[derive(Debug, Default, FromMeta)]
struct FieldOpts {
#[darling(default)]
with: Option<Ident>,
#[darling(default)]
skip_serializing: bool,
#[darling(default)]
skip_deserializing: bool,
}
const ENUM_TRANSPARENT: &str = "transparent";
const ENUM_UNION: &str = "union";
const ENUM_VARIANTS: &[&str] = &[ENUM_TRANSPARENT, ENUM_UNION];
const NO_ENUM_BEHAVIOUR_ERROR: &str = "enums require an \"enum_behaviour\" attribute, \
e.g., #[ssz(enum_behaviour = \"transparent\")]";
enum EnumBehaviour {
Transparent,
Union,
}
impl EnumBehaviour {
pub fn new(s: Option<String>) -> Option<Self> {
s.map(|s| match s.as_ref() {
ENUM_TRANSPARENT => EnumBehaviour::Transparent,
ENUM_UNION => EnumBehaviour::Union,
other => panic!(
"{} is an invalid enum_behaviour, use either {:?}",
other, ENUM_VARIANTS
),
})
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
}
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
fn parse_ssz_fields(struct_data: &syn::DataStruct) -> Vec<(&syn::Type, &syn::Ident, FieldOpts)> {
struct_data
.fields
.iter()
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
.map(|field| {
let ty = &field.ty;
let ident = match &field.ident {
Some(ref ident) => ident,
_ => panic!("ssz_derive only supports named struct fields."),
};
let field_opts_candidates = field
.attrs
.iter()
.filter(|attr| attr.path.get_ident().map_or(false, |ident| *ident == "ssz"))
.collect::<Vec<_>>();
if field_opts_candidates.len() > 1 {
panic!("more than one field-level \"ssz\" attribute provided")
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
let field_opts = field_opts_candidates
.first()
.map(|attr| {
let meta = attr.parse_meta().unwrap();
FieldOpts::from_meta(&meta).unwrap()
})
.unwrap_or_default();
(ty, ident, field_opts)
})
.collect()
}
/// Implements `ssz::Encode` for some `struct` or `enum`.
#[proc_macro_derive(Encode, attributes(ssz))]
pub fn ssz_encode_derive(input: TokenStream) -> TokenStream {
let item = parse_macro_input!(input as DeriveInput);
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
let opts = StructOpts::from_derive_input(&item).unwrap();
let enum_opt = EnumBehaviour::new(opts.enum_behaviour);
match &item.data {
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
syn::Data::Struct(s) => {
if enum_opt.is_some() {
panic!("enum_behaviour is invalid for structs");
}
ssz_encode_derive_struct(&item, s)
}
syn::Data::Enum(s) => match enum_opt.expect(NO_ENUM_BEHAVIOUR_ERROR) {
EnumBehaviour::Transparent => ssz_encode_derive_enum_transparent(&item, s),
EnumBehaviour::Union => ssz_encode_derive_enum_union(&item, s),
},
_ => panic!("ssz_derive only supports structs and enums"),
}
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
/// Derive `ssz::Encode` for a struct.
///
/// Fields are encoded in the order they are defined.
///
/// ## Field attributes
///
/// - `#[ssz(skip_serializing)]`: the field will not be serialized.
fn ssz_encode_derive_struct(derive_input: &DeriveInput, struct_data: &DataStruct) -> TokenStream {
let name = &derive_input.ident;
let (impl_generics, ty_generics, where_clause) = &derive_input.generics.split_for_impl();
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
let field_is_ssz_fixed_len = &mut vec![];
let field_fixed_len = &mut vec![];
let field_ssz_bytes_len = &mut vec![];
let field_encoder_append = &mut vec![];
for (ty, ident, field_opts) in parse_ssz_fields(struct_data) {
if field_opts.skip_serializing {
continue;
}
if let Some(module) = field_opts.with {
let module = quote! { #module::encode };
field_is_ssz_fixed_len.push(quote! { #module::is_ssz_fixed_len() });
field_fixed_len.push(quote! { #module::ssz_fixed_len() });
field_ssz_bytes_len.push(quote! { #module::ssz_bytes_len(&self.#ident) });
field_encoder_append.push(quote! {
encoder.append_parameterized(
#module::is_ssz_fixed_len(),
|buf| #module::ssz_append(&self.#ident, buf)
)
});
} else {
field_is_ssz_fixed_len.push(quote! { <#ty as ssz::Encode>::is_ssz_fixed_len() });
field_fixed_len.push(quote! { <#ty as ssz::Encode>::ssz_fixed_len() });
field_ssz_bytes_len.push(quote! { self.#ident.ssz_bytes_len() });
field_encoder_append.push(quote! { encoder.append(&self.#ident) });
}
}
let output = quote! {
impl #impl_generics ssz::Encode for #name #ty_generics #where_clause {
fn is_ssz_fixed_len() -> bool {
#(
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
#field_is_ssz_fixed_len &&
)*
true
}
fn ssz_fixed_len() -> usize {
if <Self as ssz::Encode>::is_ssz_fixed_len() {
let mut len: usize = 0;
#(
len = len
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
.checked_add(#field_fixed_len)
.expect("encode ssz_fixed_len length overflow");
)*
len
} else {
ssz::BYTES_PER_LENGTH_OFFSET
}
}
fn ssz_bytes_len(&self) -> usize {
if <Self as ssz::Encode>::is_ssz_fixed_len() {
<Self as ssz::Encode>::ssz_fixed_len()
} else {
let mut len: usize = 0;
#(
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
if #field_is_ssz_fixed_len {
len = len
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
.checked_add(#field_fixed_len)
.expect("encode ssz_bytes_len length overflow");
} else {
len = len
.checked_add(ssz::BYTES_PER_LENGTH_OFFSET)
.expect("encode ssz_bytes_len length overflow for offset");
len = len
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
.checked_add(#field_ssz_bytes_len)
.expect("encode ssz_bytes_len length overflow for bytes");
}
)*
len
}
}
2019-05-05 23:26:58 +00:00
fn ssz_append(&self, buf: &mut Vec<u8>) {
let mut offset: usize = 0;
#(
offset = offset
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
.checked_add(#field_fixed_len)
.expect("encode ssz_append offset overflow");
)*
2019-05-05 23:26:58 +00:00
2019-05-05 23:58:31 +00:00
let mut encoder = ssz::SszEncoder::container(buf, offset);
#(
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
#field_encoder_append;
)*
2019-05-05 23:58:31 +00:00
encoder.finalize();
}
}
};
output.into()
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
/// Derive `ssz::Encode` for an enum in the "transparent" method.
///
/// The "transparent" method is distinct from the "union" method specified in the SSZ specification.
/// When using "transparent", the enum will be ignored and the contained field will be serialized as
/// if the enum does not exist. Since an union variant "selector" is not serialized, it is not
/// possible to reliably decode an enum that is serialized transparently.
///
/// ## Limitations
///
/// Only supports:
/// - Enums with a single field per variant, where
/// - All fields are variably sized from an SSZ-perspective (not fixed size).
///
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
/// ## Panics
///
/// Will panic at compile-time if the single field requirement isn't met, but will panic *at run
/// time* if the variable-size requirement isn't met.
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
fn ssz_encode_derive_enum_transparent(
derive_input: &DeriveInput,
enum_data: &DataEnum,
) -> TokenStream {
let name = &derive_input.ident;
let (impl_generics, ty_generics, where_clause) = &derive_input.generics.split_for_impl();
let (patterns, assert_exprs): (Vec<_>, Vec<_>) = enum_data
.variants
.iter()
.map(|variant| {
let variant_name = &variant.ident;
if variant.fields.len() != 1 {
panic!("ssz::Encode can only be derived for enums with 1 field per variant");
}
let pattern = quote! {
#name::#variant_name(ref inner)
};
let ty = &(&variant.fields).into_iter().next().unwrap().ty;
let type_assert = quote! {
!<#ty as ssz::Encode>::is_ssz_fixed_len()
};
(pattern, type_assert)
})
.unzip();
let output = quote! {
impl #impl_generics ssz::Encode for #name #ty_generics #where_clause {
fn is_ssz_fixed_len() -> bool {
assert!(
#(
#assert_exprs &&
)* true,
"not all enum variants are variably-sized"
);
false
}
fn ssz_bytes_len(&self) -> usize {
match self {
#(
#patterns => inner.ssz_bytes_len(),
)*
}
}
fn ssz_append(&self, buf: &mut Vec<u8>) {
match self {
#(
#patterns => inner.ssz_append(buf),
)*
}
}
}
};
output.into()
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
/// Derive `ssz::Encode` for an `enum` following the "union" SSZ spec.
///
/// The union selector will be determined based upon the order in which the enum variants are
/// defined. E.g., the top-most variant in the enum will have a selector of `0`, the variant
/// beneath it will have a selector of `1` and so on.
///
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
/// # Limitations
///
/// Only supports enums where each variant has a single field.
fn ssz_encode_derive_enum_union(derive_input: &DeriveInput, enum_data: &DataEnum) -> TokenStream {
let name = &derive_input.ident;
let (impl_generics, ty_generics, where_clause) = &derive_input.generics.split_for_impl();
let patterns: Vec<_> = enum_data
.variants
.iter()
.map(|variant| {
let variant_name = &variant.ident;
if variant.fields.len() != 1 {
panic!("ssz::Encode can only be derived for enums with 1 field per variant");
}
let pattern = quote! {
#name::#variant_name(ref inner)
};
pattern
})
.collect();
let union_selectors = compute_union_selectors(patterns.len());
let output = quote! {
impl #impl_generics ssz::Encode for #name #ty_generics #where_clause {
fn is_ssz_fixed_len() -> bool {
false
}
fn ssz_bytes_len(&self) -> usize {
match self {
#(
#patterns => inner
.ssz_bytes_len()
.checked_add(1)
.expect("encoded length must be less than usize::max_value"),
)*
}
}
fn ssz_append(&self, buf: &mut Vec<u8>) {
match self {
#(
#patterns => {
let union_selector: u8 = #union_selectors;
debug_assert!(union_selector <= ssz::MAX_UNION_SELECTOR);
buf.push(union_selector);
inner.ssz_append(buf)
},
)*
}
}
}
};
output.into()
}
/// Derive `ssz::Decode` for a struct or enum.
#[proc_macro_derive(Decode, attributes(ssz))]
pub fn ssz_decode_derive(input: TokenStream) -> TokenStream {
let item = parse_macro_input!(input as DeriveInput);
let opts = StructOpts::from_derive_input(&item).unwrap();
let enum_opt = EnumBehaviour::new(opts.enum_behaviour);
match &item.data {
syn::Data::Struct(s) => {
if enum_opt.is_some() {
panic!("enum_behaviour is invalid for structs");
}
ssz_decode_derive_struct(&item, s)
}
syn::Data::Enum(s) => match enum_opt.expect(NO_ENUM_BEHAVIOUR_ERROR) {
EnumBehaviour::Transparent => panic!(
"Decode cannot be derived for enum_behaviour \"{}\", only \"{}\" is valid.",
ENUM_TRANSPARENT, ENUM_UNION
),
EnumBehaviour::Union => ssz_decode_derive_enum_union(&item, s),
},
_ => panic!("ssz_derive only supports structs and enums"),
}
}
/// Implements `ssz::Decode` for some `struct`.
2019-02-19 07:43:09 +00:00
///
/// Fields are decoded in the order they are defined.
///
/// ## Field attributes
///
/// - `#[ssz(skip_deserializing)]`: during de-serialization the field will be instantiated from a
/// `Default` implementation. The decoder will assume that the field was not serialized at all
/// (e.g., if it has been serialized, an error will be raised instead of `Default` overriding it).
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
fn ssz_decode_derive_struct(item: &DeriveInput, struct_data: &DataStruct) -> TokenStream {
let name = &item.ident;
let (impl_generics, ty_generics, where_clause) = &item.generics.split_for_impl();
let mut register_types = vec![];
let mut field_names = vec![];
let mut fixed_decodes = vec![];
let mut decodes = vec![];
let mut is_fixed_lens = vec![];
let mut fixed_lens = vec![];
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
for (ty, ident, field_opts) in parse_ssz_fields(struct_data) {
field_names.push(quote! {
#ident
});
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
// Field should not be deserialized; use a `Default` impl to instantiate.
if field_opts.skip_deserializing {
decodes.push(quote! {
let #ident = <_>::default();
});
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
fixed_decodes.push(quote! {
let #ident = <_>::default();
});
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
continue;
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
let is_ssz_fixed_len;
let ssz_fixed_len;
let from_ssz_bytes;
if let Some(module) = field_opts.with {
let module = quote! { #module::decode };
is_ssz_fixed_len = quote! { #module::is_ssz_fixed_len() };
ssz_fixed_len = quote! { #module::ssz_fixed_len() };
from_ssz_bytes = quote! { #module::from_ssz_bytes(slice) };
register_types.push(quote! {
builder.register_type_parameterized(#is_ssz_fixed_len, #ssz_fixed_len)?;
});
decodes.push(quote! {
let #ident = decoder.decode_next_with(|slice| #module::from_ssz_bytes(slice))?;
});
} else {
is_ssz_fixed_len = quote! { <#ty as ssz::Decode>::is_ssz_fixed_len() };
ssz_fixed_len = quote! { <#ty as ssz::Decode>::ssz_fixed_len() };
from_ssz_bytes = quote! { <#ty as ssz::Decode>::from_ssz_bytes(slice) };
register_types.push(quote! {
builder.register_type::<#ty>()?;
});
decodes.push(quote! {
let #ident = decoder.decode_next()?;
});
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
fixed_decodes.push(quote! {
let #ident = {
start = end;
end = end
.checked_add(#ssz_fixed_len)
.ok_or_else(|| ssz::DecodeError::OutOfBoundsByte {
i: usize::max_value()
})?;
let slice = bytes.get(start..end)
.ok_or_else(|| ssz::DecodeError::InvalidByteLength {
len: bytes.len(),
expected: end
})?;
#from_ssz_bytes?
};
});
is_fixed_lens.push(is_ssz_fixed_len);
fixed_lens.push(ssz_fixed_len);
}
let output = quote! {
impl #impl_generics ssz::Decode for #name #ty_generics #where_clause {
fn is_ssz_fixed_len() -> bool {
#(
#is_fixed_lens &&
)*
true
}
fn ssz_fixed_len() -> usize {
if <Self as ssz::Decode>::is_ssz_fixed_len() {
let mut len: usize = 0;
#(
len = len
.checked_add(#fixed_lens)
.expect("decode ssz_fixed_len overflow");
)*
len
} else {
ssz::BYTES_PER_LENGTH_OFFSET
}
}
2019-11-24 01:00:49 +00:00
fn from_ssz_bytes(bytes: &[u8]) -> std::result::Result<Self, ssz::DecodeError> {
if <Self as ssz::Decode>::is_ssz_fixed_len() {
if bytes.len() != <Self as ssz::Decode>::ssz_fixed_len() {
return Err(ssz::DecodeError::InvalidByteLength {
len: bytes.len(),
expected: <Self as ssz::Decode>::ssz_fixed_len(),
});
}
let mut start: usize = 0;
let mut end = start;
#(
#fixed_decodes
)*
Ok(Self {
#(
#field_names,
)*
})
} else {
let mut builder = ssz::SszDecoderBuilder::new(bytes);
#(
#register_types
)*
let mut decoder = builder.build()?;
#(
#decodes
)*
Ok(Self {
#(
#field_names,
)*
})
}
}
}
};
output.into()
}
Implement SSZ union type (#2579) ## Issue Addressed NA ## Proposed Changes Implements the "union" type from the SSZ spec for `ssz`, `ssz_derive`, `tree_hash` and `tree_hash_derive` so it may be derived for `enums`: https://github.com/ethereum/consensus-specs/blob/v1.1.0-beta.3/ssz/simple-serialize.md#union The union type is required for the merge, since the `Transaction` type is defined as a single-variant union `Union[OpaqueTransaction]`. ### Crate Updates This PR will (hopefully) cause CI to publish new versions for the following crates: - `eth2_ssz_derive`: `0.2.1` -> `0.3.0` - `eth2_ssz`: `0.3.0` -> `0.4.0` - `eth2_ssz_types`: `0.2.0` -> `0.2.1` - `tree_hash`: `0.3.0` -> `0.4.0` - `tree_hash_derive`: `0.3.0` -> `0.4.0` These these crates depend on each other, I've had to add a workspace-level `[patch]` for these crates. A follow-up PR will need to remove this patch, ones the new versions are published. ### Union Behaviors We already had SSZ `Encode` and `TreeHash` derive for enums, however it just did a "transparent" pass-through of the inner value. Since the "union" decoding from the spec is in conflict with the transparent method, I've required that all `enum` have exactly one of the following enum-level attributes: #### SSZ - `#[ssz(enum_behaviour = "union")]` - matches the spec used for the merge - `#[ssz(enum_behaviour = "transparent")]` - maintains existing functionality - not supported for `Decode` (never was) #### TreeHash - `#[tree_hash(enum_behaviour = "union")]` - matches the spec used for the merge - `#[tree_hash(enum_behaviour = "transparent")]` - maintains existing functionality This means that we can maintain the existing transparent behaviour, but all existing users will get a compile-time error until they explicitly opt-in to being transparent. ### Legacy Option Encoding Before this PR, we already had a union-esque encoding for `Option<T>`. However, this was with the *old* SSZ spec where the union selector was 4 bytes. During merge specification, the spec was changed to use 1 byte for the selector. Whilst the 4-byte `Option` encoding was never used in the spec, we used it in our database. Writing a migrate script for all occurrences of `Option` in the database would be painful, especially since it's used in the `CommitteeCache`. To avoid the migrate script, I added a serde-esque `#[ssz(with = "module")]` field-level attribute to `ssz_derive` so that we can opt into the 4-byte encoding on a field-by-field basis. The `ssz::legacy::four_byte_impl!` macro allows a one-liner to define the module required for the `#[ssz(with = "module")]` for some `Option<T> where T: Encode + Decode`. Notably, **I have removed `Encode` and `Decode` impls for `Option`**. I've done this to force a break on downstream users. Like I mentioned, `Option` isn't used in the spec so I don't think it'll be *that* annoying. I think it's nicer than quietly having two different union implementations or quietly breaking the existing `Option` impl. ### Crate Publish Ordering I've modified the order in which CI publishes crates to ensure that we don't publish a crate without ensuring we already published a crate that it depends upon. ## TODO - [ ] Queue a follow-up `[patch]`-removing PR.
2021-09-25 05:58:36 +00:00
/// Derive `ssz::Decode` for an `enum` following the "union" SSZ spec.
fn ssz_decode_derive_enum_union(derive_input: &DeriveInput, enum_data: &DataEnum) -> TokenStream {
let name = &derive_input.ident;
let (impl_generics, ty_generics, where_clause) = &derive_input.generics.split_for_impl();
let (constructors, var_types): (Vec<_>, Vec<_>) = enum_data
.variants
.iter()
.map(|variant| {
let variant_name = &variant.ident;
if variant.fields.len() != 1 {
panic!("ssz::Encode can only be derived for enums with 1 field per variant");
}
let constructor = quote! {
#name::#variant_name
};
let ty = &(&variant.fields).into_iter().next().unwrap().ty;
(constructor, ty)
})
.unzip();
let union_selectors = compute_union_selectors(constructors.len());
let output = quote! {
impl #impl_generics ssz::Decode for #name #ty_generics #where_clause {
fn is_ssz_fixed_len() -> bool {
false
}
fn from_ssz_bytes(bytes: &[u8]) -> Result<Self, ssz::DecodeError> {
// Sanity check to ensure the definition here does not drift from the one defined in
// `ssz`.
debug_assert_eq!(#MAX_UNION_SELECTOR, ssz::MAX_UNION_SELECTOR);
let (selector, body) = ssz::split_union_bytes(bytes)?;
match selector.into() {
#(
#union_selectors => {
<#var_types as ssz::Decode>::from_ssz_bytes(body).map(#constructors)
},
)*
other => Err(ssz::DecodeError::UnionSelectorInvalid(other))
}
}
}
};
output.into()
}
fn compute_union_selectors(num_variants: usize) -> Vec<u8> {
let union_selectors = (0..num_variants)
.map(|i| {
i.try_into()
.expect("union selector exceeds u8::max_value, union has too many variants")
})
.collect::<Vec<u8>>();
let highest_selector = union_selectors
.last()
.copied()
.expect("0-variant union is not permitted");
assert!(
highest_selector <= MAX_UNION_SELECTOR,
"union selector {} exceeds limit of {}, enum has too many variants",
highest_selector,
MAX_UNION_SELECTOR
);
union_selectors
}