docs: hoist ADR status (#1407)

Addresses celestiaorg/celestia-app#1161 (comment) ## Questions 1. Do we want to make the same change across other celestiaorg repos? 2. Should we consider an org wide ADR template? 1. celestiaorg/celestia-app#1415
celestia-org · Feb 22, 2023 · 026e868 · 026e868
1 parent e550ece
commit 026e868
Show file tree

Hide file tree

Showing 12 changed files with 95 additions and 104 deletions.
diff --git a/docs/architecture/adr-001-abci++-adoption.md b/docs/architecture/adr-001-abci++-adoption.md
@@ -1,5 +1,9 @@
 # ADR 001: ABCI++ Adoption
 
+## Status
+
+Implemented
+
 ## Changelog
 
 - 2022-03-03: Initial Commit
@@ -32,10 +36,6 @@ While the adoption of ABCI++ is inevitable given the already made decision by up
 - [Alternatives for Message Inclusion.](https://github.com/celestiaorg/celestia-app/blob/92341dd68ee6e555ec6c0bb780afa3a1c8243a93/adrs/adr008:adopt-ABC%2B%2B-early.md#alternative-approaches)
 - [Alternatives for Picking a square size.](https://github.com/celestiaorg/celestia-core/issues/454)
 
-## Decision
-
-Proposed and initial implementation is complete.
-
 ## Detailed Design
 
 ### [#631](https://github.com/celestiaorg/celestia-core/pull/631) Simplified version of ABCI++ (`celestia-core`)
@@ -370,10 +370,6 @@ func (app *App) ProcessProposal(req abci.RequestProcessProposal) abci.ResponsePr
 }
 ```
 
-## Status
-
-The proposed and initial implementation is complete.
-
 ## Consequences
 
 ### Positive

diff --git a/docs/architecture/adr-002-qgb-valset.md b/docs/architecture/adr-002-qgb-valset.md
@@ -1,5 +1,9 @@
 # ADR 002: QGB ValSet
 
+## Status
+
+Implemented
+
 ## Context
 
 To accommodate the requirements of the [Quantum Gravity Bridge](https://github.com/celestiaorg/quantum-gravity-bridge/blob/76efeca0be1a17d32ef633c0fdbd3c8f5e4cc53f/src/QuantumGravityBridge.sol), We will need to add support for `ValSet`s, i.e. Validator Sets, which reflect the current state of the bridge validators.
@@ -328,11 +332,3 @@ ctx.EventManager().EmitEvent(
     ),
 )
 ```
-
-## Status
-
-Accepted
-
-## References
-
-- {reference link}
diff --git a/docs/architecture/adr-003-qgb-data-commitments.md b/docs/architecture/adr-003-qgb-data-commitments.md
@@ -1,5 +1,9 @@
 # ADR 003: QGB Data Commitments
 
+## Status
+
+Implemented
+
 ## Context
 
 To accommodate the requirements of the [Quantum Gravity Bridge](https://github.com/celestiaorg/quantum-gravity-bridge/blob/76efeca0be1a17d32ef633c0fdbd3c8f5e4cc53f/src/QuantumGravityBridge.sol), We will need to add support for `DataCommitment`s messages, i.e. commitments generated over a set of blocks to attest their existence.
@@ -129,11 +133,3 @@ ctx.EventManager().EmitEvent(
     ),
 )
 ```
-
-## Status
-
-Accepted
-
-## References
-
-- {reference link}
diff --git a/docs/architecture/adr-004-qgb-relayer-security.md b/docs/architecture/adr-004-qgb-relayer-security.md
@@ -1,5 +1,9 @@
 # ADR 004: QGB Relayer Security
 
+## Status
+
+Implemented
+
 ## Changelog
 
 - 2022-06-05: Synchronous QGB implementation
@@ -397,10 +401,6 @@ In a nutshell, a new valset will be emitted if any of the following is true:
 significantPowerDiff = intCurrMembers.PowerDiff(*intLatestMembers) > 0.05
 ```
 
-## Status
-
-Accepted
-
 ## References
 
 - Tracker issue for the tasks [here](https://github.com/celestiaorg/celestia-app/issues/467).
diff --git a/docs/architecture/adr-005-qgb-reduce-state-usage.md b/docs/architecture/adr-005-qgb-reduce-state-usage.md
@@ -1,5 +1,9 @@
 # ADR 005: QGB Reduce State Usage
 
+## Status
+
+Deprecated in favor of [orchestrator-relayer#65](https://github.com/celestiaorg/orchestrator-relayer/pull/66)
+
 ## Context
 
 The first design for the QGB was to use the state extensively to store all the QGB-related data: Attestations, `Valset Confirms` and `DataCommitment Confirms`.
@@ -40,7 +44,7 @@ However, slashing will be very difficult, especially for liveness, i.e. an orche
 
 Remove the `MsgValsetConfirm` defined in [here](https://github.com/celestiaorg/celestia-app/blob/a965914b8a467f0384b17d9a8a0bb1ac62f384db/proto/qgb/msgs.proto#L24-L49)
 And also, the `MsgDataCommitmentConfirm` defined in [here](
-https://github.com/celestiaorg/celestia-app/blob/a965914b8a467f0384b17d9a8a0bb1ac62f384db/proto/qgb/msgs.proto#L55-L76).
+<https://github.com/celestiaorg/celestia-app/blob/a965914b8a467f0384b17d9a8a0bb1ac62f384db/proto/qgb/msgs.proto#L55-L76>).
 Which were the way orchestrators were able to post confirms to the QGB module.
 Then, keep only the state that is created in [EndBlocker](https://github.com/celestiaorg/celestia-app/blob/a965914b8a467f0384b17d9a8a0bb1ac62f384db/x/qgb/abci.go#L12-L16).
 Which are `Attestations`, i.e. `Valset`s and `DataCommitmentRequest`s.
@@ -63,7 +67,7 @@ We will need to decide on two things:
 ## Detailed Design
 
 The proposed design consists of keeping the same transaction types we currently have : the `MsgValsetConfirm` defined in [here](https://github.com/celestiaorg/celestia-app/blob/a965914b8a467f0384b17d9a8a0bb1ac62f384db/proto/qgb/msgs.proto#L24-L49), and the `MsgDataCommitmentConfirm` defined in [here](
-https://github.com/celestiaorg/celestia-app/blob/a965914b8a467f0384b17d9a8a0bb1ac62f384db/proto/qgb/msgs.proto#L55-L76). However, remove  all the message server checks defined in the [msg_server.go](https://github.com/celestiaorg/celestia-app/blob/9867b653b2a253ba01cb7889e2dbfa6c9ff67909/x/qgb/keeper/msg_server.go) :
+<https://github.com/celestiaorg/celestia-app/blob/a965914b8a467f0384b17d9a8a0bb1ac62f384db/proto/qgb/msgs.proto#L55-L76>). However, remove  all the message server checks defined in the [msg_server.go](https://github.com/celestiaorg/celestia-app/blob/9867b653b2a253ba01cb7889e2dbfa6c9ff67909/x/qgb/keeper/msg_server.go) :
 
 ```go
 // ValsetConfirm handles MsgValsetConfirm.
@@ -95,10 +99,6 @@ For posting transactions, we will rely on gas fees as a mechanism to limit malic
 
 When it comes to slashing, we can add the `dataRoot` of the blocks to the state during `ProcessProposal`,  `FinalizeCommit`, or in some other way to be defined. Then, we will have a way to slash orchestrators after a certain period of time if they didn't post any confirms. The exact details of this will be left for another ADR.
 
-## Status
-
-Proposed
-
 ## Consequences
 
 ### Positive

diff --git a/docs/architecture/adr-006-non-interactive-defaults.md b/docs/architecture/adr-006-non-interactive-defaults.md
@@ -1,5 +1,9 @@
 # ADR 006: Non-interactive Defaults, Wrapped Transactions, and Subtree Root Message Inclusion Checks
 
+## Status
+
+Implemented
+
 > **Note**
 > Unlike normal tendermint/cosmos ADRs, this ADR isn't for deciding whether or not we will implement non-interactive defaults. The goal of this document is to help reviewers and future readers understand what non-interactive defaults are, the considerations that went into the initial implementation, and how it differs from the original specs.
 
@@ -678,10 +682,6 @@ func (app *App) ProcessProposal(req abci.RequestProcessProposal) abci.ResponsePr
 
 The current implementation performs many different estimation and calculation steps. It might be possible to amortize these calculations to each transaction, which would make it a lot easier to confidently arrange an optimal block.
 
-## Status
-
-Accepted
-
 ## Consequences
 
 ### Positive

diff --git a/docs/architecture/adr-007-universal-share-prefix.md b/docs/architecture/adr-007-universal-share-prefix.md
@@ -1,5 +1,9 @@
 # ADR 007: Universal Share Prefix
 
+## Status
+
+Implemented
+
 ## Terminology
 
 - **compact share**: a type of share that can accommodate multiple units. Currently, compact shares are used for transactions, and evidence is to efficiently pack this information into as few shares as possible.
@@ -120,10 +124,6 @@ Logic
     1. All shares contain a share version that belongs to a list of supported versions (initially this list contains version `0`)
     1. All shares in a reserved namespace belong to one share sequence
 
-## Status
-
-Implemented
-
 ## Consequences
 
 ### Positive

diff --git a/docs/architecture/adr-008-square-size-independent-message-commitments.md b/docs/architecture/adr-008-square-size-independent-message-commitments.md
@@ -1,5 +1,9 @@
 # ADR 008: square size independent message commitments
 
+## Status
+
+Implemented
+
 ## Changelog
 
 - 03.08.2022: Initial Draft
@@ -63,16 +67,12 @@ func MinSquareSize(shareCount uint64) uint64 {
 }
 ```
 
-## Status
-
-Implemented
-
 ## Consequences
 
 ### Negative
 
 1. The amount of subtree roots per commitment is O(sqrt(n)), while n is the number of message shares. The worst case for the number of subtree roots is depicted in the diagram below - an entire block missing one share.
-  ![Interactive Commitment 2](./assets/complexity.png)  
+  ![Interactive Commitment 2](./assets/complexity.png)
 The worst case for the current implementation depends on the square size. If it is the worst square size, as in `msgMinSquareSize`, it is O(sqrt(n)) as well. On the other hand, if the message is only in one row, then it is O(log(n)).
 Therefore the height of the tree over the subtree roots is in this implementation O(log(sqrt(n))), while n is the number of message shares. In the current implementation, it varies from O(log(sqrt(n))) to O(log(log(n))) depending on the square size.
 

diff --git a/docs/architecture/adr-009-non-interactive-default-rules-for-reduced-padding.md b/docs/architecture/adr-009-non-interactive-default-rules-for-reduced-padding.md
@@ -1,5 +1,9 @@
 # ADR 009: Non-Interactive Default Rules for Reduced Padding
 
+## Status
+
+Implemented
+
 ## Changelog
 
 - 14.11.2022: Initial Draft
@@ -18,7 +22,7 @@ Proposed
 
 The upside of this proposal is that it reduces the inter-message padding. The downside is that a message inclusion proof will not be as efficient for large square sizes so the proof will be larger.
 
-> **Note**  
+> **Note**
 > This analysis assumes the implementation of [celestia-app#1004](https://github.com/celestiaorg/celestia-app/issues/1004). If the tree over the subtree roots is not a Namespace Merkle Tree then both methods have the same proof size.
 
 As an example, take the diagram below. Message 1 is 3 shares long and message 2 is 11 shares long.
@@ -119,17 +123,17 @@ Each row consists of one subtree root, which means if you have log(n) rows you w
 
 ![Current ni rules proof size](./assets/current-ni-rules-proof-size.png)
 
-NMT-Node size := 32 bytes + 2\*8 bytes = 48 bytes  
+NMT-Node size := 32 bytes + 2\*8 bytes = 48 bytes
 MT-Node size := 32 bytes
 
-Proof size = subtree roots (rows) + subtree roots (last row) + blue nodes (parity shares) + 2 \* blue nodes (`DataRoot`)  
-Proof size = (log(n) + log(k) + log(n)) \* NMT-Node size  + 2\*log(k) \* MT-Node size  
+Proof size = subtree roots (rows) + subtree roots (last row) + blue nodes (parity shares) + 2 \* blue nodes (`DataRoot`)
+Proof size = (log(n) + log(k) + log(n)) \* NMT-Node size  + 2\*log(k) \* MT-Node size
 Proof size = 48 \* (2\*log(n) + log(k)) + 64 \*log(k)
 
 ### Current Non-Interactive Default Rules for k/4
 
-Proof size = subtree roots (rows) + subtree roots (last row) + blue nodes (parity shares) + 2 \* blue nodes (`DataRoot`)  
-Proof size = (k/4 + log(k) + k/4) \* NMT-Node size  + 2\*log(k) \* MT-Node size  
+Proof size = subtree roots (rows) + subtree roots (last row) + blue nodes (parity shares) + 2 \* blue nodes (`DataRoot`)
+Proof size = (k/4 + log(k) + k/4) \* NMT-Node size  + 2\*log(k) \* MT-Node size
 Proof size = 48 \* (k/2 + log(k)) + 64 \*log(k)
 
 ### Proposed Non-Interactive Default Rules
@@ -138,14 +142,14 @@ Each row consists of sqrt(n)/log(n) subtree roots. Which makes in total sqrt(n)
 
 ![Proposed ni rules proof size](./assets/proposed-ni-rules-proof-size.png)
 
-Proof size = subtree roots (all rows) + subtree roots (last row) + blue nodes (parity shares) + 2 \* blue nodes (`DataRoot`)  
-Proof size = (sqrt(n) + log(k) + log(n)) \* NMT-Node size  + 2\*log(k) \* MT-Node size  
+Proof size = subtree roots (all rows) + subtree roots (last row) + blue nodes (parity shares) + 2 \* blue nodes (`DataRoot`)
+Proof size = (sqrt(n) + log(k) + log(n)) \* NMT-Node size  + 2\*log(k) \* MT-Node size
 Proof size = 48 \* (sqrt(n) + log(k) + log(n)) + 64 \*log(k)
 
 ### Proposed Non-Interactive Default Rules for k/4
 
-Proof size = subtree roots (rows) + subtree roots (last row) + blue nodes (parity shares) + 2 \* blue nodes (`DataRoot`)  
-Proof size = (**k/2** + log(k) + k/4) \* NMT-Node size  + 2\*log(k) \* MT-Node size  
+Proof size = subtree roots (rows) + subtree roots (last row) + blue nodes (parity shares) + 2 \* blue nodes (`DataRoot`)
+Proof size = (**k/2** + log(k) + k/4) \* NMT-Node size  + 2\*log(k) \* MT-Node size
 Proof size = 48 \* (3k/4 + log(k)) + 64 \*log(k)
 
 ## 5. What is the worst constructible block with the most amount of padding with old and new non-interactive default rules?
@@ -192,10 +196,6 @@ The worst-case padding decreases from 1.1 GB to 0.8 GB in 2 GB Blocks. In the cu
 
 You can further optimize the proof size by using the fact the Namespace is known and the same for all the subtree roots. You can do the same trick for parity shares as the namespace is fixed for them too. Both of these optimizations are not included in the analysis and would save the bytes that are used to store the namespace.
 
-## Status
-
-Accepted
-
 ## Consequences
 
 ### Positive