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

Revert "Change enum values to be powers of two (fix #3417)" #4604

Closed
wants to merge 2 commits into from

Conversation

intelliot
Copy link
Collaborator

@intelliot intelliot commented Jun 30, 2023

Consider the function conditionMet, specifically lines 121 and 122 and 146-147 of Handler.h:

1c2ae10#diff-41d8d23b15cb894b075204dcd88d7bf9ca223fa77f43e28bb22cf38b805cabfcL121-L122

1c2ae10#diff-41d8d23b15cb894b075204dcd88d7bf9ca223fa77f43e28bb22cf38b805cabfcL146-L147

The change is broken: where it previously, it checked for a specific condition (e.g. NEEDS_CLOSED_LEDGER) and only entered the if block if it was met, it now executes that code if any condition is set.

A prime example of a RPC request that fails is the tx command, which has NEEDS_NETWORK_CONNECTION and will now not work unless the server has both a current and a closed ledger.

The problem isn't the powers of two change. It's how the checks where restructured.

Originally posted by @nbougalis in #4601 (comment)

Reverts #4239

@intelliot intelliot requested review from nbougalis and ckeshava June 30, 2023 18:33
@ckeshava
Copy link
Collaborator

ckeshava commented Jul 3, 2023

Yes, reverting the commit will fix the problem. But I feel we also need to add additional unit tests, so that such blindspots are covered for the future.

@intelliot intelliot requested a review from HowardHinnant July 4, 2023 00:14
@intelliot intelliot added the Bug label Jul 4, 2023
@intelliot intelliot added this to the 1.12 milestone Jul 4, 2023
@intelliot
Copy link
Collaborator Author

This is ready to merge because it just reverts PR #4239, and the original author of that PR has approved.

@intelliot intelliot requested a review from ximinez July 5, 2023 15:50
Copy link
Contributor

@HowardHinnant HowardHinnant left a comment

Choose a reason for hiding this comment

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

This revertible code isn't readable because it checks like a bit-mask even though it is not a bit-mask.

If the power-of-2 fix doesn't have the right behavior, it should be fixed with code that is more readable than this revert results in. E.g. change the checks to use == instead of &.

That being said, @ximinez has an analysis that says the behavior between these two versions is the same, in which case I strongly prefer the power-of-2 version.

@ximinez
Copy link
Collaborator

ximinez commented Jul 5, 2023

Veto. Copying my comment from #4601 (comment):

A prime example of a RPC request that fails is the tx command, which has NEEDS_NETWORK_CONNECTION and will now not work unless the server has both a current and a closed ledger.

I don't think there's a behavior change at all. I commented on the original PR to that effect: #4239 (comment)

Here's more detail.

In the old code,

NO_CONDITION == 0
NEEDS_NETWORK_CONNECTION == 1 == 0b0001
NEEDS_CURRENT_LEDGER     == 3 == 0b0011
NEEDS_CLOSED_LEDGER      == 5 == 0b0101

Consider the condition on line 110:

condition_required & NEEDS_NETWORK_CONNECTION

which is equivalent to

condition_required & 0b0001

condition_required can be any of those 4 values.

condition_required condition_required & NEEDS_NETWORK_CONNECTION boolean result
NO_CONDITION == 0 0 false
NEEDS_NETWORK_CONNECTION == 0b0001 0b0001 true
NEEDS_CURRENT_LEDGER == 0b0011 0b0001 true
NEEDS_CLOSED_LEDGER == 0b0101 0b0001 true

Basically, for any value except NO_CONDITION (0), the expression evaluates to true.

The new values are 0, 1, 2, and 4, respectively, but the result is the same. condition_required != NO_CONDITION in the new code is functionally equivalent to condition_required & NEEDS_NETWORK_CONNECTION in the old code.

It should be clear that the results are the same for the other values.

@intelliot
Copy link
Collaborator Author

Closing because @ckeshava and @ximinez agree that this should not be reverted; see #4601 (comment)

@intelliot intelliot closed this Jul 12, 2023
@intelliot intelliot deleted the revert-4239-issue3417 branch August 12, 2023 16:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
Status: 📋 Backlog
Development

Successfully merging this pull request may close these issues.

5 participants