-
Notifications
You must be signed in to change notification settings - Fork 41
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[DRAFT] WIP QML Load Snapshot Signet (160,000 height) #424
base: main
Are you sure you want to change the base?
Conversation
cb409e3
to
670ca05
Compare
With cb409e3 rebased and got rid off the extra options_mode settings which are not relevant for this evaluation. Also deleted the extra loadsnapshot.cpp file and instead added the load snapshot function in src/node/interfaces.cpp |
Should I be getting a binary specific for this issue? I tried loading a signet snapshot into the one for #421 but was unable to... |
670ca05
to
b77f7c1
Compare
With b77f7c1 minor tweaks to the src/node/interfaces.cpp snapshotload() function. It mimics the rpc call loadtxoutset from bitcoin/bitcoin repository Next will connect the loading to the progress bar |
b77f7c1
to
c59e557
Compare
with c59e557 Integrated the UI with the C++ backend, implementing a progress callback function to dynamically update the loading progress based on the number of coins processed. Touched src/validation.h and src/validation.cpp to extract the coin loading progress. Next: will update the "Loaded Snapshot" stack with the real info from the snapshot |
c59e557
to
e3f8c70
Compare
With e3f8c70 moved the load snapshot functionality from nodeModel to optionsModel since it make it easier to store the snapshot info and reduces redundant variable declarations. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
With e3f8c70 moved the load snapshot functionality from nodeModel to optionsModel since it make it easier to store the snapshot info and reduces redundant variable declarations.
e3f8c70 is a regression because optionsmodel is not an appropriate place for this, and we are now violating the role of settings.json to store the state of a process.
NodeModel and AppMode are more appropriate places for this information. But the entire approach here is still a concept NACK as we are performing unecessary changes to validation and chainparams.
src/qml/models/options_model.cpp
Outdated
if (new_snapshot_load_completed != m_snapshot_load_completed) { | ||
m_snapshot_load_completed = new_snapshot_load_completed; | ||
if (m_onboarded) { | ||
m_node.updateRwSetting("snapshotloadcompleted", new_snapshot_load_completed); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This would be polluting settings.json that stores read/write configuration options for core -> with a state of a process. We cannot do this.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Makes sense, although I have a question for when we would want to allow the user to load a snapshot during onboarding, could the snapshot file path be stored in the settings.json?
Ok I tested artefact#235 and functionality & UI seems to work. Didn't see block clock so couldn't verify easily. Adding some UI feedback below:
|
e3f8c70
to
90dc0dd
Compare
with 90dc0dd rebased over main and addressed @jarolrod concerns by reverting back to just a minimal integration of the loading snapshot functionality. @yashrajd I will address the UI issues next As a follow up will investigate how to access the snapshot verification progress Also looking into how to make the snapshot data persist and be available for display (i.e. snapshot block height, date, and merkle hash) |
For the chainparams the changes there are only temporary for evaluating and testing within this repo. I checked upstream and the signet snapshot has already been hardcoded so once our repo is synced the changes in chainparams.cpp can be discarded. |
90dc0dd
to
85592f5
Compare
with 85592f5 made the snapshot activation persist on node restart by making use of the interface with chain->hasAssumedValidChain Also changed the green-check icon size to 30 pix Next looking into how to extract the progress from the ActivateSnapshot process |
85592f5
to
955c65d
Compare
with 955c65d modified the snapshotLoad() function in src/node/interfaces.cpp and the initializeSnapshot() function in nodemodel.cpp to extract progress information using regular expressions. This design decision aimed to keep snapshot loading mechanisms self-contained. However, it looks like the use of Perhaps the way to go is to use the handler in src/node/interfaces_ui ... will explore that path then update |
955c65d
to
0eba513
Compare
with 0eba513 reverted back to minimal snapshot load functionality. Snapshot loading progress is still not were I want it to be so in the meantime replaced the progress bar with placeholder text. Also updated the snapshot info display. once successfully loaded, and I'm using the hardcoded info in |
0eba513
to
cd2ec69
Compare
With cd2ec69 added a section to keep track of snapshot loading, it is not ideal so will look for better ways to achieve it. Since the core functionality is there will ask for "pre-reviews" before un-drafting |
5d0756c
to
262bb03
Compare
with 5d0756c updated with 262bb03 updated Next looking into how to finesse the snapshot loading progress... |
262bb03
to
a6bede5
Compare
with a6bede5 removed comments from |
a6bede5
to
799b4cd
Compare
with 799b4cd updated Also wanted to see if the Win64 CI completed successfully, which it did. Next will ask for feedback before un-drafting |
799b4cd
to
f23ffb8
Compare
f23ffb8
to
5d2ffb0
Compare
with 5d2ffb0 updated snapshotLoad() in src/node/interfaces.cpp. Left the polling loop for now as I investigate how to handle loading the snapshot if Pre-syncing and Syncing headers is still happening. Also removed JS functions and moved the logic to the nodemodel |
This introduce the UI flow to load a AssumeUTXO snapshot into the Bitcoin Core App. It modifies the connection seetings and adds a SnapshotSettings file, Icon, and modified profress bar. It has been rebased
Adds minimal wiring to connect QML GUI to loading a signet UTXO snapshot via the connection settings. Uses SnapshotSettings.qml to allow user interaction. Modifies src/interfaces/node.h, src/node/interfaces.cpp and chainparams.cpp (temporarily for signet snapshot testing) to expose snapshot loading functionality through the node model. Current limitations: - Not integrated with onboarding process - Requires manual navigation to connection settings after initial startup - Snapshot verification progress is not implemented yet Testing: 1. Start the node 2. Complete onboarding 3. Navigate to connection settings 4. Load snapshot from provided interface
5d2ffb0
to
8beb50a
Compare
8beb50a rebased over main |
@@ -371,6 +371,13 @@ class SigNetParams : public CChainParams { | |||
|
|||
vFixedSeeds.clear(); | |||
|
|||
m_assumeutxo_data = MapAssumeutxo{ | |||
{ | |||
160000, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
8beb50a: the signet snapshot was merged into Bitcoin Core more than a year ago in bitcoin/bitcoin@edbed31. This also seems to be using the older serialization format from before bitcoin/bitcoin#29612.
I guess that's all fine for a proof of concept, but it would be better if the main branch of this repo was more up to date.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hi Sjors! Thanks for taking a look and yes completely agree with:
it would be better if the main branch of this repo was more up to date
We are currently taking steps to sync with upstream, meanwhile I wanted to get the ball rolling, although I will pull in the update versions of the assume utxo functionality so that when the sync happens this PR can be merged with minimal refactoring.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Another thing you could do is look into making a PR to the Bitcoin Core repo with your proposed interface changes. One way to go about that is to somehow use them in existing or a new RPC method. E.g. snapshot load progress is probably useful to have available.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Another thing you could do is look into making a PR to the Bitcoin Core repo with your proposed interface changes.
That's the current medium term plan, once we have an implementation that works and makes sense here we plan to go upstream (bitcoin/bitcoin) and open a PR there.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A handler similar to handleNotifyBlockTip will be needed to so the UI interface can be event driven instead of polling the interfaces module.
// Wait for the block to appear in the block index | ||
//TODO: remove this once another method is implemented | ||
constexpr int max_wait_seconds = 600; // 10 minutes | ||
for (int i = 0; i < max_wait_seconds; ++i) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We shouldn't have a blocking component here. If there are any issues we should just fail.
ArgsManager& args() { return *Assert(Assert(m_context)->args); } | ||
ChainstateManager& chainman() { return *Assert(m_context->chainman); } | ||
NodeContext* m_context{nullptr}; | ||
std::atomic<double> m_snapshot_progress{0.0}; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I imagine the actual progress should be owned by the object in charge of doing the processing. ChainStateManager?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes! Will set it up so that the handler notifies the snapshot progress and that ownership belongs to ChainstateManager.
descriptionSize: 17 | ||
descriptionLineHeight: 1.1 | ||
description: snapshotInfo && snapshotInfo["date"] ? | ||
qsTr("It contains transactions up to %1. Newer transactions still need to be downloaded." + |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's not really "transactions up to", it's all coins (utxo set) that exist at the snapshot height.
Maybe say something like:
It contains a snapshot as of block %1. Once loaded, all blocks from %2 to the present are download. After that Bitcoin Core is ready to use, while all historical blocks from genesis to %2 are downloaded and validated in the background.
(intentionally leaves it vague what's in the snapshot)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In the sequence, this is the fourth snapshot-related screen (state) people see:
- Snapshot option in settings where people learn about the benefit of this feature
- "Load snapshot" screen that gives more detail about how it works
- "Loading snapshot" screen state with a loader that tells people they can do other stuff in the mean time
- "Snapshot loaded" screen state that informs people what was imported, what to keep in mind and what will happen next (this is the screen that we are talking about in this thread)
I think the point that not all historical transactions are in the snapshot is worth clarifying. The first time people are exposed to what this feature does is in 2, so I'd probably start tweaking at that point. We call it a "recent transaction snapshot" there, which is more precise than just "transactions" ("unspent transaction snapshot" would be a little closer).
We don't need to be 100% technically accurate and I'm not sure users are well-served if we're completely vague about this either. They should be able to form a good enough mental model for making decisions and understanding their impact.
So how about this?
It contains recent transactions up to %1. Newer and historical transactions will be automatically downloaded and verified.
Or this?
It contains unspent transactions up to %1. Newer and historical transactions will be automatically downloaded and verified.
Either way, we should ensure to tweak the text across the sequence if we make changes here.
We show the block height in a "View details" option below (did not check if it's implemented in this PR already), so not sure whether we need to include that in the text as well.
As for what happens next, my thinking was that we just say that the data will be filled in and verified here (keep it short), and then have indicators contextually in the UI as to what is happening (balance, send and transaction list have respective load states). So less absorption of logic here, and more contextual info in the appropriate places.
Sorry, this has gotten quite long. Hope it's a helpful response.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We don't need to be 100% technically accurate and I'm not sure users are well-served if we're completely vague about this either.
I think it's better to be incomplete than incorrect. The nice thing about being vague is that it's clear to the reader there's more to learn.
One reason the phrasing "recent / historical transactions" is potentially confusing is that people may think they can recover historical wallet transactions with it. But it has nothing to do with that.
The concept that's useful to explain is that the snapshot allows us to start syncing at block %1, saving time by skipping historical blocks. This way the user can get started (with their wallet) earlier. Meanwhile it's safe, because we do check those historical blocks in the background later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
TL;DR I reviewed the onboarding text again, and it seems like the use of transactions
terminology in the snapshot section provides context continuity with what's said before...see screenshot below.
Unless we come up with something better, I'd be fine with retaining what we have currently though I agree it's inaccurate. Suggestions below:
I think it's better to be incomplete than incorrect. The nice thing about being vague is that it's clear to the reader there's more to learn.
It does seem like we can strike a better balance between technical accuracy & usefulness to user here, and use vagueness where required.
We show the block height in a "View details" option below (did not check if it's implemented in this PR already), so not sure whether we need to include that in the text as well.
Agree, this was in the initial mockups I did. edit: removed something here. read on...
The first time people are exposed to what this feature does is in 2, so I'd probably start tweaking at that point.
Agree.
- "Load snapshot" screen that gives more detail about how it works.
This screen can omit mention of transactions "The snapshot allows you to start using the application more quickly. Once loaded, it will be automatically verified in the background. Learn more" instead of "You can start using the application more quickly by loading a recent transaction snapshot. It will be automatically verified in the background."
- "Snapshot loaded" screen state that informs people what was imported, what to keep in mind and what will happen next (this is the screen that we are talking about in this thread)
I'd suggest something like "The loaded snapshot contains data up to [some date]. The initial download will continue now. The snapshot is verified in the background." instead of "It contains transactions up to September 10, 2023. Newer transactions still need to be downloaded. The data will be verified in the background."
edit: rephrased my own suggestion slightly.
Based on #421. For evaluation purposes only!
GUI Integration for UTXO Snapshot Loading
Overview
This PR adds initial GUI support for loading a
signet
UTXO snapshot at height 160,000, building on Bitcoin Core'sassumeutxo
infrastructure.What This PR Does
Implementation Details
Core Components Modified
src/node/interfaces.cpp
)snapshotLoad()
based on theloadtxoutset
RPCsrc/qml/models/nodemodel.cpp
)src/qml/components/SnapshotSettings.qml
)Key Design Decisions
assumeutxo
changesTesting Instructions
Ubuntu 22.04 Screenshots
Expected Behavior
Future Work
Handler
interface for snapshot eventsassumeutxo
changestestnet
andmainnet
)Notes for Reviewers
Snapshot Compatibility
signet
UTXO snapshotm_assumeutxo_data
(the changes there can be discarded once synced since
m_assumeutxo_data
is already hardcoded)This is a work in progress - feedback welcome on the approach and implementation details.