Skip to content

Latest commit

 

History

History
399 lines (320 loc) · 14.8 KB

MIGRATING.md

File metadata and controls

399 lines (320 loc) · 14.8 KB

Migrating

This guide lists API changes between cw-plus major releases.

v0.13.x -> v0.14.0

Breaking Issues / PRs

  • MultiIndex index fn params now include the pk #670

The idx_fn param of the MultiIndex constructor now has signature fn(&[u8], &T) -> IK, where the first param is the primary key of the index (in raw format), and the second param is the associated value. That allows us to use the pk or parts of it for building the index key.

Migration of existing code is straight-forward. Just add an (unused) _pk param to the index function definition:

fn build_map<'a>() -> IndexedMap<'a, &'a str, Data, DataIndexes<'a>> {
   let indexes = DataIndexes {
-     name: MultiIndex::new(|d| d.name.clone(), "data", "data__name"),
+     name: MultiIndex::new(|_pk, d| d.name.clone(), "data", "data__name"),
      age: UniqueIndex::new(|d| d.age, "data__age"),
      name_lastname: UniqueIndex::new(

If you want to leverage this new functionality, take a look at the pk_based_index() test / example in src/indexed_map.rs.

v0.11.0 -> v0.12.0

Breaking Issues / PRs

  • Type safe Bounds #462

Bounds are now type-safe. That means the bound type must match the key / sub-key you are ranging over.

Migration code example:

fn list_allowed(
    deps: Deps,
    start_after: Option<String>,
    limit: Option<u32>,
) -> StdResult<ListAllowedResponse> {
     let limit = limit.unwrap_or(DEFAULT_LIMIT).min(MAX_LIMIT) as usize;
-    let start = match start_after {
-        Some(x) => Some(Bound::exclusive(deps.api.addr_validate(&x)?.into_string())),
-        None => None,
-    };
+    let addr = maybe_addr(deps.api, start_after)?;
+    let start = addr.as_ref().map(Bound::exclusive);

     let allow = ALLOW_LIST
         .range(deps.storage, start, None, Order::Ascending)

Here the ALLOW_LIST key is of type &Addr. That's why we use as_ref() before the map() that builds the bound. Notice also that this "forces" you to use addr_validate(), in order to build a bound over the proper type.

You can still use untyped bounds, with the ExclusiveRaw and InclusiveRaw enum types. Migration code example, in case you want to keep your raw bounds:

pub fn query_all_allowances(
    deps: Deps,
    env: Env,
    start_after: Option<String>,
    limit: Option<u32>,
) -> StdResult<AllAllowancesResponse> {
     let limit = calc_limit(limit);
     // we use raw addresses here....
-    let start = start_after.map(Bound::exclusive);
+    let start = start_after.map(|s| Bound::ExclusiveRaw(s.into()));

     let allowances = ALLOWANCES
         .range(deps.storage, start, None, Order::Ascending)

Notice that here we build a bound for an address, and using a raw bound allows us to skip address validation / build up.

See storage-plus README.md for more information on Bound.

v0.10.3 -> v0.11.0

Breaking Issues / PRs

  • Incorrect I32Key Index Ordering #489 / Signed int keys order #582

As part of range iterators revamping, we fixed the order of signed integer keys. You shouldn't change anything in your code base for this, but if you were using signed keys and relying on their ordering, that has now changed for the better. Take into account also that the internal representation of signed integer keys has changed. So, if you have data stored under signed integer keys you would need to migrate it, or recreate it under the new representation.

As part of this, a couple helpers for handling int keys serialization and deserialization were introduced:

  • from_cw_bytes Integer (signed and unsigned) values deserialization.
  • to_cw_bytes - Integer (signed and unsigned) values serialization.

You shouldn't need these, except when manually handling raw integer keys serialization / deserialization.

Migration code example:

#[cfg_attr(not(feature = "library"), entry_point)]
pub fn migrate(deps: DepsMut, _env: Env, _msg: Empty) -> Result<Response, ContractError> {
  let version: Version = CONTRACT_VERSION.parse()?;
  let storage_version: Version = get_contract_version(deps.storage)?.version.parse()?;

  if storage_version < version {
    set_contract_version(deps.storage, CONTRACT_NAME, CONTRACT_VERSION)?;

    // Do the migration
    // Original map
    let signed_int_map: Map<IntKeyOld<i8>, String> = Map::new("signed_int_map");

    // New map (using a different namespace for safety. It could be the same with enough care)
    let signed_int_map_new: Map<i8, String> = Map::new("signed_int_map-v2");

    // Obtain all current keys (this will need to be paginated if there are many entries,
    // i.e. i32 or i64 instead of i8).
    // This may be gas intensive
    let current = signed_int_map
            .range(deps.storage, None, None, Order::Ascending)
            .collect::<StdResult<Vec<_>>>()?;

    // Store length for quality control (adjust if paginated)
    let current_count = current.len();

    // Remove the old map keys
    for (k, _) in current.iter() {
      signed_int_map.remove(deps.storage, (*k).into());
    }

    // Save in new format
    for (k, v) in current.into_iter() {
      signed_int_map_new.save(deps.storage, k, &v)?;
    }

    // Confirm old map is empty
    if signed_int_map
            .keys_raw(deps.storage, None, None, Order::Ascending)
            .next()
            .is_some()
    {
      return Err(StdError::generic_err("Original still not empty!").into());
    }

    // Obtain new keys, and confirm their amount.
    // May be gas intensive.
    let new_count = signed_int_map_new
            .keys_raw(deps.storage, None, None, Order::Ascending)
            .count();

    if current_count != new_count {
      return Err(StdError::generic_err(format!(
        "Current ({}) and new ({}) counts differ!",
        current_count, new_count
      ))
              .into());
    }
  }

  Ok(Response::new())
}

  • Rename cw0 to utils #471 / Cw0 rename #508

The cw0 package was renamed to cw-utils. The required changes are straightforward:

diff --git a/contracts/cw1-subkeys/Cargo.toml b/contracts/cw1-subkeys/Cargo.toml
index 1924b655..37af477d 100644
--- a/contracts/cw1-subkeys/Cargo.toml
+++ b/contracts/cw1-subkeys/Cargo.toml
-cw0 = { path = "../../packages/cw0", version = "0.10.3" }
+cw-utils = { path = "../../packages/utils", version = "0.10.3" }
diff --git a/contracts/cw1-subkeys/src/contract.rs b/contracts/cw1-subkeys/src/contract.rs
index b4852225..f20a65ec 100644
--- a/contracts/cw1-subkeys/src/contract.rs
+++ b/contracts/cw1-subkeys/src/contract.rs
-use cw0::Expiration;
+use cw_utils::Expiration;

  • Deprecate range to range_raw #460 / range to range raw #576.

range was renamed to range_raw (no key deserialization), and range_de to range. This means you can now count on the key to be deserialized for you, which results in cleaner / simpler code.

There are some examples in the contracts code base, by example:

diff --git a/contracts/cw3-fixed-multisig/src/contract.rs b/contracts/cw3-fixed-multisig/src/contract.rs
index 48a60083..2f2ef70d 100644
--- a/contracts/cw3-fixed-multisig/src/contract.rs
+++ b/contracts/cw3-fixed-multisig/src/contract.rs
@@ -385,21 +384,20 @@ fn list_votes(
         .range(deps.storage, start, None, Order::Ascending)
         .take(limit)
         .map(|item| {
-            let (key, ballot) = item?;
-            Ok(VoteInfo {
-                voter: String::from_utf8(key)?,
+            item.map(|(addr, ballot)| VoteInfo {
+                voter: addr.into(),
                 vote: ballot.vote,
                 weight: ballot.weight,
             })

If you don't need key deserialization, just can just rename range to range_raw and you are good to go.

If you want key deserialization for indexes, you need to specify the primary key type as a last (optional) argument in the index key specification. If not specified, it defaults to (), which means that primary keys will not only not be deserialized, but also not provided. This is for backwards-compatibility with current indexes specifications, and may change in the future once these features are stabilized. See packages/storage-plus/src/indexed_map.rs tests for reference.


Renamed methods:

  • range -> range_raw
  • keys -> keys_raw
  • prefix_range -> prefix_range_raw
  • range_de -> range
  • keys_de -> keys
  • prefix_range_de -> prefix_range

Finally, this applies to all the Map-like types, including indexed maps.


  • UniqueIndex / MultiIndex key consistency #532 / Index keys consistency #568

For both UniqueIndex and MultiIndex, returned keys (deserialized or not) are now the associated primary keys of the data. This is a breaking change for MultiIndex, as the previous version returned a composite of the index key and the primary key.

Examples of required changes are:

diff --git a/packages/storage-plus/src/indexed_map.rs b/packages/storage-plus/src/indexed_map.rs
index 9f7178af..d11d501e 100644
--- a/packages/storage-plus/src/indexed_map.rs
+++ b/packages/storage-plus/src/indexed_map.rs
@@ -722,7 +722,7 @@ mod test {
             last_name: "".to_string(),
             age: 42,
         };
-        let pk1: &[u8] = b"5627";
+        let pk1: &str = "5627";
         map.save(&mut store, pk1, &data1).unwrap();

         let data2 = Data {
@@ -730,7 +730,7 @@ mod test {
             last_name: "Perez".to_string(),
             age: 13,
         };
-        let pk2: &[u8] = b"5628";
+        let pk2: &str = "5628";
         map.save(&mut store, pk2, &data2).unwrap();

         let data3 = Data {
@@ -738,7 +738,7 @@ mod test {
             last_name: "Young".to_string(),
             age: 24,
         };
-        let pk3: &[u8] = b"5629";
+        let pk3: &str = "5629";
         map.save(&mut store, pk3, &data3).unwrap();

         let data4 = Data {
@@ -746,7 +746,7 @@ mod test {
             last_name: "Bemberg".to_string(),
             age: 43,
         };
-        let pk4: &[u8] = b"5630";
+        let pk4: &str = "5630";
         map.save(&mut store, pk4, &data4).unwrap();

         let marias: Vec<_> = map
@@ -760,8 +760,8 @@ mod test {
         assert_eq!(2, count);

         // Remaining part (age) of the index keys, plus pks (bytes) (sorted by age descending)
-        assert_eq!((42, pk1.to_vec()), marias[0].0);
-        assert_eq!((24, pk3.to_vec()), marias[1].0);
+        assert_eq!(pk1, marias[0].0);
+        assert_eq!(pk3, marias[1].0);

         // Data
         assert_eq!(data1, marias[0].1);

  • Remove the primary key from the MultiIndex key specification #533 / MultiIndex primary key spec removal #569

The primary key is no longer needed when specifying MultiIndex keys. This simplifies both MultiIndex definitions and usage.

Required changes are along the lines of:

diff --git a/packages/storage-plus/src/indexed_map.rs b/packages/storage-plus/src/indexed_map.rs
index 022a4504..c7a3bb9d 100644
--- a/packages/storage-plus/src/indexed_map.rs
+++ b/packages/storage-plus/src/indexed_map.rs
@@ -295,8 +295,8 @@ mod test {
     }

     struct DataIndexes<'a> {
-        // Second arg is for storing pk
-        pub name: MultiIndex<'a, (String, String), Data, String>,
+        // Last type parameters are for signaling pk deserialization
+        pub name: MultiIndex<'a, String, Data, String>,
         pub age: UniqueIndex<'a, u32, Data, String>,
         pub name_lastname: UniqueIndex<'a, (Vec<u8>, Vec<u8>), Data, String>,
     }
@@ -326,11 +326,7 @@ mod test {
     // Can we make it easier to define this? (less wordy generic)
     fn build_map<'a>() -> IndexedMap<'a, &'a str, Data, DataIndexes<'a>> {
         let indexes = DataIndexes {
-            name: MultiIndex::new(
-                |d, k| (d.name.clone(), unsafe { String::from_utf8_unchecked(k) }),
-                "data",
-                "data__name",
-            ),
+            name: MultiIndex::new(|d| d.name.clone(), "data", "data__name"),
             age: UniqueIndex::new(|d| d.age, "data__age"),
             name_lastname: UniqueIndex::new(
                 |d| index_string_tuple(&d.name, &d.last_name),
@@ -469,8 +462,8 @@ mod test {
         // index_key() over MultiIndex works (empty pk)
         // In a MultiIndex, an index key is composed by the index and the primary key.
         // Primary key may be empty (so that to iterate over all elements that match just the index)
-        let key = ("Maria".to_string(), "".to_string());
-        // Use the index_key() helper to build the (raw) index key
+        let key = "Maria".to_string();
+        // Use the index_key() helper to build the (raw) index key with an empty pk
         let key = map.idx.name.index_key(key);
         // Iterate using a bound over the raw key
         let count = map

  • cw3-fixed-multisig requires threshold during instantiation instead of required_weight parameter

Threshold type was moved to packages/utils along with surrounding implementations like ThresholdResponse etc.

use cw_utils::Threshold;

pub struct InstantiateMsg {
    pub voters: Vec<Voter>,
-   pub required_weight: u64,
+   pub threshold: Threshold,
    pub max_voting_period: Duration,
}

Non-breaking Issues / PRs

  • Deprecate IntKey #472 / Deprecate IntKey #547

The IntKey wrapper and its type aliases are marked as deprecated, and will be removed in the next major version. The migration is straightforward, just remove the wrappers:

diff --git a/contracts/cw20-merkle-airdrop/src/contract.rs b/contracts/cw20-merkle-airdrop/src/contract.rs
index e8d5ea57..bdf555e8 100644
--- a/contracts/cw20-merkle-airdrop/src/contract.rs
+++ b/contracts/cw20-merkle-airdrop/src/contract.rs
@@ -6,7 +6,6 @@ use cosmwasm_std::{
 };
 use cw2::{get_contract_version, set_contract_version};
 use cw20::Cw20ExecuteMsg;
-use cw_storage_plus::U8Key;
 use sha2::Digest;
 use std::convert::TryInto;

@@ -113,7 +112,7 @@ pub fn execute_register_merkle_root(

     let stage = LATEST_STAGE.update(deps.storage, |stage| -> StdResult<_> { Ok(stage + 1) })?;

-    MERKLE_ROOT.save(deps.storage, U8Key::from(stage), &merkle_root)?;
+    MERKLE_ROOT.save(deps.storage, stage, &merkle_root)?;
     LATEST_STAGE.save(deps.storage, &stage)?;

     Ok(Response::new().add_attributes(vec![
  • See CHANGELOG.md for the full list of non-breaking changes.