Skip to content
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

Voting Contract: vote and close uses current status #60

Merged
merged 7 commits into from
Feb 3, 2022

Conversation

ueco-jb
Copy link

@ueco-jb ueco-jb commented Feb 1, 2022

closes #55

Closing and voting methods now use current_status() helper, which underneath can update status to Rejected.
Because of that, first time closing a proposal that should be rejected could fail, because there was check in place to see, if status... is not already rejected.
To mitigate that, I used simple solution with saving status before closing, in order to detect if someone tries to close proposal first of 1+nth time.

I added extra .save() under each update status, since in case of early exit status update would be lost. So that doesn't work, and I don't like it. :)
So someone tries to vote on proposal, update_status() changes state to Rejected and vote is rejected, but then validator queries the proposal and sees that status was the same as before - probably just Open.

I didn't change anything in state.rs file, simply moved impl Proposal so that it would be directly under its structure, instead on bottom of file for some reason.

@ueco-jb ueco-jb self-assigned this Feb 1, 2022
@ueco-jb ueco-jb force-pushed the 55-voting-contract-status-reading-bugs branch 3 times, most recently from 3c8eeda to 12c3af7 Compare February 1, 2022 11:34
…ing and closing current status is used and saved
@ueco-jb ueco-jb force-pushed the 55-voting-contract-status-reading-bugs branch from 12c3af7 to 0cdf7df Compare February 1, 2022 11:37
@ueco-jb ueco-jb marked this pull request as ready for review February 1, 2022 11:51
uint
uint previously requested changes Feb 1, 2022
packages/voting-contract/src/multitest/early_end.rs Outdated Show resolved Hide resolved
packages/voting-contract/src/lib.rs Outdated Show resolved Hide resolved
packages/voting-contract/src/lib.rs Outdated Show resolved Hide resolved
packages/voting-contract/src/lib.rs Outdated Show resolved Hide resolved
@uint
Copy link
Contributor

uint commented Feb 1, 2022

Overall: really awesome you caught these problems, but I have some comments!

I added extra .save() under each update status, since in case of early exit status update would be lost. So someone tries to vote on proposal, update_status() changes state to Rejected and vote is rejected, but then validator queries the proposal and sees that status was the same as before.

Got a test for this?

I didn't change anything in state.rs file, simply moved impl Proposal so that it would be directly under its structure, instead on bottom of file for some reason.

👌

Copy link
Contributor

@uint uint left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Unmarking this PR as request changes so as not to block it. Oops.

@uint uint self-requested a review February 1, 2022 19:23
@ueco-jb
Copy link
Author

ueco-jb commented Feb 2, 2022

Got a test for this?

I mistakenly took this "closing twice test" as proof of that, but now I see that it was due to this check-for-rejected-status mechanism I added.

This save doesn't write to the storage directly, but to a cache. All such changes are committed to persistent storage later if the fn succeeds and discarded if the fn returns an error.

This still bothers me though. Because as I wrote in description:

So someone tries to vote on proposal, update_status() changes state to Rejected and vote is rejected, but then validator queries the proposal and sees that status was the same as before - probably just Open.

I think this will be confusing and we should mitigate that, but now I don't have any idea how.
Wrong idea here - unless we start returning some empty response with error message in couple cases...?

@ueco-jb ueco-jb dismissed uint’s stale review February 2, 2022 08:28

Reviewer tried to dismiss it himself.

@ueco-jb
Copy link
Author

ueco-jb commented Feb 2, 2022

Anyway, PR solves problem from issue and works.

I referenced my problems in new issue, so this one could actually be merged.

@uint
Copy link
Contributor

uint commented Feb 2, 2022

So someone tries to vote on proposal, update_status() changes state to Rejected and vote is rejected, but then validator queries the proposal and sees that status was the same as before - probably just Open.

I think this will be confusing and we should mitigate that, but now I don't have any idea how. Wrong idea here - unless we start returning some empty response with error message in couple cases...?

Does this actually happen though? The status query doesn't check the storage directly, but uses a helper that's supposed to return the current status taking expiration into account:
https://github.com/confio/poe-contracts/blob/main/packages/voting-contract/src/lib.rs#L231

I think the idea with this stuff is that queries use current_status() and executions use update_status() and then everything checks out.

Copy link
Contributor

@maurolacy maurolacy left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good. Just some questions / comments, and what I think is an error when closing.

packages/voting-contract/src/lib.rs Outdated Show resolved Hide resolved
packages/voting-contract/src/lib.rs Outdated Show resolved Hide resolved
packages/voting-contract/src/lib.rs Show resolved Hide resolved
packages/voting-contract/src/lib.rs Outdated Show resolved Hide resolved
return Err(ContractError::Rejected {});
}

prop.update_status(&env.block);
Copy link
Contributor

@maurolacy maurolacy Feb 2, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is the key thing to do. Wouldn't this have to be done in the other cases where current_status() is being called as well? (i.e. our previous "execute proposal" fix).
That means, using current_status() only for queries, and update_status() for execution everywhere. And, "never" reading prop.status directly. More consistent and robust, I think.

We can use an update_status() impl that returns the updated status, as mentioned above (just an idea).

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think the idea with this stuff is that queries use current_status() and executions use update_status() and then everything checks out.

Yeah, you're right. I just realized, that despite status in memory not being correct, in the end it is correct for end user who uses query. We "simply" must be carefull to make sure we use that helper everywhere.

Just wrote a variant of the same above.

Copy link
Contributor

@maurolacy maurolacy Feb 2, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If the code errors below, this status update will be discarded. And if it does not error, the status will be updated and saved later anyway. So, there's really no need to call update_status instead of current_status.

The bug is: not calling any of those, and reading the status directly.

packages/voting-contract/src/lib.rs Outdated Show resolved Hide resolved
packages/voting-contract/src/state.rs Show resolved Hide resolved
@ueco-jb
Copy link
Author

ueco-jb commented Feb 2, 2022

I think the idea with this stuff is that queries use current_status() and executions use update_status() and then everything checks out.

Yeah, you're right. I just realized, that despite status in memory not being correct, in the end it is correct for end user who uses query.
We "simply" must be carefull to make sure we use that helper everywhere.

@ueco-jb ueco-jb force-pushed the 55-voting-contract-status-reading-bugs branch from 7a9de1f to 1adbde0 Compare February 2, 2022 09:20
@maurolacy
Copy link
Contributor

maurolacy commented Feb 2, 2022

We "simply" must be carefull to make sure we use that helper everywhere.

I was just thinking that a (relatively cumbersome) way to do this would be to have / use two structs, one for state, with private members, and one for serialisation / deserialization, with pub ones.

@uint
Copy link
Contributor

uint commented Feb 2, 2022

We "simply" must be carefull to make sure we use that helper everywhere.

I was just thinking that a (relatively cumbersome) way to do this would be to have / use two structs, one for state, with private members, and one for serialisation / deserialization, with pub ones.

Yeah, that's one way to do it. I opened an issue for this: #63

Copy link
Contributor

@uint uint left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@ueco-jb ueco-jb force-pushed the 55-voting-contract-status-reading-bugs branch from bbbd8ce to 94daa76 Compare February 3, 2022 08:24
@ueco-jb ueco-jb merged commit 56844bb into main Feb 3, 2022
@ueco-jb ueco-jb deleted the 55-voting-contract-status-reading-bugs branch February 3, 2022 08:36
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[voting-contract] More status issues
4 participants