feat: add std feature and almost full no_std support. (#192)

# Rationale for this change

We want `no_std` support.

# What changes are included in this PR?

* Added a `std` and `perf` feature to `proof-of-sql`.
* Made `proof-of-sql-parser` `no_std`.
* Updated dependency features to disable `std` usage where not needed.
* Added CI jobs to ensure these features always pass.

The one caveat is that lalrpop needs a patch while waiting on
lalrpop/lalrpop#953 in order to have true
`no_std` support. To test `no_std` compatibility, we can build on a
target that does not support `std`. There is a new CI job that targets
`thumbv7em-none-eabi`. This fails without the lalrpop patch.

Once `lalrpop` cuts a release, we can simply bump the version and remove
the patch in the CI.

# Are these changes tested?

Yes

.github/workflows/lint-and-test.yml

@@ -49,6 +49,19 @@ jobs:
         run: cargo check -p proof-of-sql --no-default-features --features="test"
       - name: Run cargo check (proof-of-sql) (just "blitzar" feature)
         run: cargo check -p proof-of-sql --no-default-features --features="blitzar"
+      - name: Run cargo check (proof-of-sql) (just "arrow" feature)
+        run: cargo check -p proof-of-sql --no-default-features --features="arrow"
+      - name: Run cargo check (proof-of-sql) (just "rayon" feature)
+        run: cargo check -p proof-of-sql --no-default-features --features="rayon"
+      - name: Run cargo check (proof-of-sql) (just "perf" feature)
+        run: cargo check -p proof-of-sql --no-default-features --features="perf"
+      - name: Run cargo check (proof-of-sql) (just "std" feature)
+        run: cargo check -p proof-of-sql --no-default-features --features="std"
+      - name: Run cargo check (proof-of-sql-parser) with no_std target. This requires a lalrpop patch.
+        run: |
+          rustup target add thumbv7em-none-eabi
+          printf '\n[patch.crates-io]\nlalrpop = { git = "https://github.com/lalrpop/lalrpop", rev = "173597c" }\nlalrpop-util = { git = "https://github.com/lalrpop/lalrpop", rev = "173597c" }\n' >> Cargo.toml
+          cargo check -p proof-of-sql-parser --target thumbv7em-none-eabi
 
   test:
     name: Test Suite
@@ -83,10 +96,10 @@ jobs:
         run: cargo test -p proof-of-sql --no-run --no-default-features --features="arrow"
       - name: Dry run cargo test (proof-of-sql) (blitzar feature only)
         run: cargo test -p proof-of-sql --no-run --no-default-features --features="blitzar"
-      - name: Dry run cargo test (proof-of-sql) (no features)
-        run: cargo test -p proof-of-sql --no-run --no-default-features
-      - name: Run cargo test (proof primitives - Dory) (no features - i.e. not using blitzar)
-        run: cargo test proof_primitive::dory::dory_compute_commitments_test --no-default-features
+      - name: Dry run cargo test (proof-of-sql) (std feature only)
+        run: cargo test -p proof-of-sql --no-run --no-default-features --features="std"
+      - name: Run cargo test (proof primitives - Dory) (std feature only - i.e. not using blitzar)
+        run: cargo test proof_primitive::dory::dory_compute_commitments_test --no-default-features --features="std"
       - name: Run hello_world example
         run: cargo run --example hello_world --features="blitzar test"
       - name: Run posql_db example

Cargo.toml

@@ -10,26 +10,26 @@ version = "0.0.0" # DO NOT CHANGE THIS LINE! This will be automatically updated
 license-file = "LICENSE"
 
 [workspace.dependencies]
-ahash = { version = "0.8.11", default-features = false, features = ["runtime-rng"] }
+ahash = { version = "0.8.11", default-features = false }
 # alloy-primitives = { version = "0.8.1" }
 # alloy-sol-types = { version = "0.8.1" }
 ark-bls12-381 = { version = "0.4.0" }
 ark-curve25519 = { version = "0.4.0" }
-ark-ec = { version = "0.4.0", features = [ "parallel" ] }
-ark-ff = { version = "0.4.0", features = [ "parallel" ] }
-ark-poly = { version = "0.4.0", features = [ "parallel" ] }
+ark-ec = { version = "0.4.0" }
+ark-ff = { version = "0.4.0" }
+ark-poly = { version = "0.4.0" }
 ark-serialize = { version = "0.4.0" }
-ark-std = { version = "0.4.0", features = [ "parallel" ] }
+ark-std = { version = "0.4.0", default-features = false }
 arrayvec = { version = "0.7", default-features = false }
 arrow = { version = "51.0" }
 arrow-csv = { version = "51.0" }
 bit-iter = { version = "1.1.1" }
 bigdecimal = { version = "0.4.5", default-features = false, features = ["serde"] }
-blake3 = { version = "1.3.3" }
+blake3 = { version = "1.3.3", default-features = false }
 blitzar = { version = "3.1.0" }
 bumpalo = { version = "3.11.0" }
 bytemuck = {version = "1.16.3", features = ["derive"]}
-byte-slice-cast = { version = "1.2.1" }
+byte-slice-cast = { version = "1.2.1", default-features = false }
 clap = { version = "4.5.4" }
 criterion = { version = "0.5.1" }
 chrono = { version = "0.4.38", default-features = false }
@@ -38,12 +38,11 @@ derive_more = { version = "0.99" }
 flexbuffers = { version = "2.0.0" }
 # forge-script = { git = "https://github.com/foundry-rs/foundry", tag = "nightly-bf1a39980532f76cd76fd87ee32661180f606435" }
 indexmap = { version = "2.1", default-features = false }
-itertools = { version = "0.13.0" }
+itertools = { version = "0.13.0", default-features = false, features = ["use_alloc"] }
 lalrpop = { version = "0.21.0" }
-lalrpop-util = { version = "0.20.0", default-features = false }
-lazy_static = { version = "1.4.0" }
+lalrpop-util = { version = "0.21.0", default-features = false }
 merlin = { version = "2" }
-num-traits = { version = "0.2" }
+num-traits = { version = "0.2", default-features = false }
 num-bigint = { version = "0.4.4", default-features = false }
 opentelemetry = { version = "0.23.0" }
 opentelemetry-jaeger = { version = "0.20.0" }
@@ -54,11 +53,11 @@ rand = { version = "0.8", default-features = false }
 rand_core = { version = "0.6", default-features = false }
 rayon = { version = "1.5" }
 serde = { version = "1", default-features = false }
-serde_json = { version = "1" }
-snafu = { version = "0.8.4", default-features = false, features = ["std"] }
+serde_json = { version = "1", default-features = false, features = ["alloc"] }
+snafu = { version = "0.8.4", default-features = false }
 tiny-keccak = { version = "2.0.2", features = [ "keccak" ] }
 # tokio = { version = "1.39.3" }
-tracing = { version = "0.1.36" }
+tracing = { version = "0.1.36", default-features = false }
 tracing-opentelemetry = { version = "0.22.0" }
 tracing-subscriber = { version = "0.3.0" }
 wasm-bindgen = { version = "0.2.92" }

crates/proof-of-sql-parser/src/lib.rs

@@ -1,4 +1,5 @@
 #![doc = include_str!("../README.md")]
+#![no_std]
 extern crate alloc;
 
 /// Module for handling an intermediate decimal type received from the lexer.

crates/proof-of-sql/Cargo.toml

@@ -35,7 +35,6 @@ chrono = { workspace = true, features = ["serde"] }
 derive_more = { workspace = true }
 indexmap = { workspace = true, features = ["serde"] }
 itertools = { workspace = true }
-lazy_static = { workspace = true }
 merlin = { workspace = true, optional = true }
 num-traits = { workspace = true }
 num-bigint = { workspace = true, default-features = false }
@@ -74,10 +73,13 @@ flexbuffers = { workspace = true }
 development = ["arrow-csv"]
 
 [features]
-default = ["arrow", "blitzar", "rayon"]
-arrow = ["dep:arrow"]
-blitzar = ["dep:blitzar", "dep:merlin"]
-test = ["dep:rand"]
+default = ["arrow", "perf"]
+arrow = ["dep:arrow", "std"]
+blitzar = ["dep:blitzar", "dep:merlin", "std"]
+test = ["dep:rand", "std"]
+perf = ["blitzar", "rayon", "ark-ec/parallel", "ark-poly/parallel", "ark-ff/asm"]
+rayon = ["dep:rayon", "std"]
+std = ["snafu/std"]
 
 [lints]
 workspace = true

crates/proof-of-sql/src/base/commitment/mod.rs

@@ -1,5 +1,6 @@
 //! Types for creation and utilization of cryptographic commitments to proof-of-sql data.
 use crate::base::scalar::Scalar;
+use alloc::vec::Vec;
 #[cfg(feature = "blitzar")]
 pub use blitzar::{
     compute::{init_backend, init_backend_with_config, BackendConfig},

crates/proof-of-sql/src/lib.rs

@@ -1,4 +1,5 @@
 #![doc = include_str!("../README.md")]
+#![cfg_attr(not(feature = "std"), no_std)]
 extern crate alloc;
 
 pub mod base;

crates/proof-of-sql/src/sql/parse/query_expr_tests.rs

@@ -2,7 +2,7 @@ use super::ConversionError;
 use crate::{
     base::{
         database::{ColumnType, TableRef, TestSchemaAccessor},
-        map::{indexmap, IndexMap},
+        map::{indexmap, IndexMap, IndexSet},
     },
     sql::{
         parse::QueryExpr,
@@ -1881,7 +1881,10 @@ fn select_group_and_order_by_preserve_the_column_order_reference() {
     let (t, accessor) = get_test_accessor();
     let base_cols: [&str; N] = ["i", "i0", "i1", "s"]; // sorted because of `select: [cols = ... ]`
     let base_ordering = [Asc, Desc, Asc, Desc];
-    for (idx, perm_cols) in base_cols.into_iter().permutations(N).unique().enumerate() {
+    for (idx, perm_cols) in IndexSet::from_iter(base_cols.into_iter().permutations(N))
+        .into_iter()
+        .enumerate()
+    {
         let perm_col_plans = perm_cols
             .iter()
             .sorted()

crates/proof-of-sql/src/sql/postprocessing/group_by_postprocessing.rs

@@ -12,7 +12,6 @@ use proof_of_sql_parser::{
     Identifier,
 };
 use serde::{Deserialize, Serialize};
-use std::collections::HashMap;
 
 /// A group by expression
 #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
@@ -158,7 +157,7 @@ impl GroupByPostprocessing {
                 )
             })
             .collect::<PostprocessingResult<Vec<AliasedResultExpr>>>()?;
-        let group_by_identifiers = by_ids.into_iter().unique().collect();
+        let group_by_identifiers = Vec::from_iter(IndexSet::from_iter(by_ids));
         Ok(Self {
             remainder_exprs,
             group_by_identifiers,
@@ -190,14 +189,22 @@ impl<S: Scalar> PostprocessingStep<S> for GroupByPostprocessing {
     fn apply(&self, owned_table: OwnedTable<S>) -> PostprocessingResult<OwnedTable<S>> {
         // First evaluate all the aggregated columns
         let alloc = Bump::new();
-        let evaluated_columns: HashMap<AggregationOperator, Vec<(Identifier, OwnedColumn<S>)>> =
-            self.aggregation_exprs
-                .iter()
-                .map(|(agg_op, expr, id)| -> PostprocessingResult<_> {
-                    let evaluated_owned_column = owned_table.evaluate(expr)?;
-                    Ok((*agg_op, (*id, evaluated_owned_column)))
-                })
-                .process_results(|iter| iter.into_group_map())?;
+        let evaluated_columns = self
+            .aggregation_exprs
+            .iter()
+            .map(|(agg_op, expr, id)| -> PostprocessingResult<_> {
+                let evaluated_owned_column = owned_table.evaluate(expr)?;
+                Ok((*agg_op, (*id, evaluated_owned_column)))
+            })
+            .process_results(|iter| {
+                iter.fold(
+                    IndexMap::<_, Vec<_>>::default(),
+                    |mut lookup, (key, val)| {
+                        lookup.entry(key).or_default().push(val);
+                        lookup
+                    },
+                )
+            })?;
         // Next actually do the GROUP BY
         let group_by_ins = self
             .group_by_identifiers