- Apr 23, 2024
-
-
Bastian Köcher authored
-
- Apr 17, 2024
-
-
Sergej Sakac authored
This PR introduces changes enabling the transfer of coretime regions via XCM. TL;DR: There are two primary issues that are resolved in this PR: 1. The `mint` and `burn` functions were not implemented for coretime regions. These operations are essential for moving assets to and from the XCM holding register. 2. The transfer of non-fungible assets through XCM was previously disallowed. This was due to incorrectly benchmarking non-fungible asset transfers via XCM, which led to assigning it a weight of `Weight::Max`, effectively preventing its execution. ### `mint_into` and `burn` implementation This PR addresses the issue with cross-chain transferring regions back to the Coretime chain. Remote reserve transfers are performed by withdrawing and depositing the asset to and from the holding registry. This requires the asset to support burning and minting functionality. This PR adds burning and minting; however, they work a bit differently than usual so that the associated region record is not lost when burning. Instead of removing all the data, burning will set the owner of the region to `None`, and when minting it back, it will set it to an actual value. So, when cross-chain transferring, withdrawing into the registry will remove the region from its original owner, and when depositing it from the registry, it will set its owner to another account This was originally implemented in this PR: #3455, however we decided to move all of it to this single PR (https://github.com/paritytech/polkadot-sdk/pull/3455#discussion_r1547324892) ### Fixes made in this PR - Update the `XcmReserveTransferFilter` on coretime chain since it is meant as a reserve chain for coretime regions. - Update the XCM benchmark to use `AssetTransactor` instead of assuming `pallet-balances` for fungible transfers. - Update the XCM benchmark to properly measure weight consumption for nonfungible reserve asset transfers. ATM reserve transfers via the extrinsic do not work since the weight for it is set to `Weight::max()`. Closes: https://github.com/paritytech/polkadot-sdk/issues/865 --------- Co-authored-by:
Branislav Kontur <[email protected]> Co-authored-by:
Francisco Aguirre <[email protected]> Co-authored-by:
Dónal Murray <[email protected]>
-
- Apr 16, 2024
-
-
Oliver Tale-Yazdi authoredChanges: - Saturate in the input validation of he drop history function or pallet-broker. --------- Signed-off-by:Oliver Tale-Yazdi <[email protected]>
-
- Apr 15, 2024
-
-
Dónal Murray authored
The first test proves that parachains who were migrated over on a legacy lease can renew without downtime. The exception is if their lease expires in period 0 - aka within `region_length` timeslices after `start_sales` is called. The second test is designed such that it passes if the issue exists and should be fixed. This will require an intervention on Kusama to add these renewals to storage as it is too tight to schedule a runtime upgrade before the start_sales call. All leases will still have at least two full regions of coretime.
-
Bastian Köcher authored
Part of: https://github.com/paritytech/polkadot-sdk/issues/4107
-
- Apr 05, 2024
-
-
Dónal Murray authored
This PR adds a check that `max_timeslices > 0` and errors if not. It also adds a test for this behaviour and cleans up some misleading docs.
-
- Feb 08, 2024
-
-
Dónal Murray authored
Leases can be force set, but since `Leases` is a `StorageValue`, if a lease misses its sale rotation in which it should expire, it can never be cleared. This can happen if a lease is added with an `until` timeslice that lies in a region whose sale has already started or has passed, even if the timeslice itself hasn't passed. This solves that issue in a minimal way, with all expired leases being cleaned up in each sale rotation, not just the ones that are expiring in the coming region. TODO: - [x] Write test
-
- Jan 27, 2024
-
-
Sergej Sakac authored
This PR allows Coretime regions to be transferable via XCM. --------- Co-authored-by:Dónal Murray <[email protected]>
-
- Dec 29, 2023
-
-
Sergej Sakac authored
With the current code, when a user interlaces their region, the end result will be three regions in the state: - the non-interlaced region - first part of the interlaced region - second part of the interlaced region The existing implementation retains the non-interlaced region in the state, leading to a problematic scenario: 1. User 1 acquires a region from the market. 2. User 1 then interlaces this region. 3. Subsequently, User 1 transfers one part of the interlaced regions to User 2. Despite this transfer, User 1 retains the ability to assign the entire original non-interlaced region, which is inconsistent with the fact that they no longer own one of the interlaced parts. This PR resolves the issue by removing the original region, ensuring that only the two new interlaced regions remain in the state.
-
- Sep 18, 2023
-
-
Sergej Sakac authored
This PR includes the following fix: - [x] The `duration` is always set to zero in the `RegionDropped` event. This is fixed in this PR. Also added some additional tests to cover some cases that aren't covered : - [x] Selling a partitioned region to the instantaneous coretime pool. - [x] Partitioning a region after assigning it to a particular task. - [x] Interlacing a region after assigning it to a particular task.
-
- Aug 24, 2023
-
-
Gavin Wood authored
* Add Broker pallet * Flesh out CorePart * Repotting and fleshing out * more drafting * process timeslice * Test Fungibles completed * Auctions * Price morphing * First tests * Tidying up config/status * Docs * Timeslice todying * More Timeslice tidying * Tests] * Repotting. * Tests * Tests * System InstaPool cores and payout * Better Relay Test framework * Tests and instapool fixes * Support NFT interface * Proper renewals * Better events, results * Test transfer * Renewal test * Repot some impls and make dispatchables. * Better weight * Test migration * Document events * Introduce durations * Core count * Allow reassignment * Better naming * Error docs * Docs * Formatting * Advance notice period is in RC blocks, not timeslices * Docs * Formatting * Docs * Missing file * Added some events * Events for all dispatchables * Remove benchmark * Fix * Adds benchmark for configure and some basic setup * Adds benchmark for reserve and unreserve * Adds a couple of more benchmarks * Docs * Event * Fix * Adds benchmark for purchase * Dedup * Add some weight breakdowns * Repotting * Adds more benchmarks * Renaming and one more event * Sale event * Better price API and docs * Avoid possibility of clobbering renewal record * Avoid possibility of clobbering renewal record * Fixes a few benchmarks * Another test * More tests * Drop history test * Rename and CORE_MASK_BITS constant * Update frame/broker/src/dispatchable_impls.rs Co-authored-by:
Nikhil Gupta <[email protected]> Co-authored-by:
-
