Rework backtracking and state management #62
Merged
+150,760
−65
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Closed
pradyunsg
approved these changes
Nov 27, 2020
pfmoore
approved these changes
Nov 27, 2020
There was a problem hiding this comment.
I agree the comments make the logic a lot easier to follow. The new process looks good to me, I like the fact that each round the invariant is that the top stack item is in an unpinned state, that makes the reasoning much easier to follow.
The new structure gives me much more confidence that the logic is correct 🙂
The "forward" part of resolution now works like this: * A "root state" is populated. This state does not produce a "pin" but only populate the root requirements, so subsequence rounds can use it as the basis. * Each subsequent round pushes a state, pin a candidate for it, and check whether the pin results in a conflict. * If no conflcits occur, the round ends, and a new round is started. When a conflict happens, the backtracking process discards the current round's state, and try to modify the previous state so the conflict can be avoided. If this modification is unsuccessful, the still previous state is tried, until a workable state is found (or we run out of things to try). This makes each round and its associating state's lifeline easier to reason with. Each successful forward round would create a state with a pin in the stack. A round ending with backtracking will not result in a new state in the stack, but have a previously-pinned state unpinned (and modified), with everything newer than it removed.
uranusjr
force-pushed
the
backtrack-redo
branch
from
935353d to
c218d1d
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
For pypa/pip#9011.
I broke down the change into several commits. The last commit is the real "meat", others are only clean ups and test cases.
Quoting from the commit message:
The "forward" part of resolution now works like this:
When a conflict happen, the backtracking process discards the current round's state, and try to modify the previous state so the conflict can be avoided. If this modification is unsuccessful, the still previous
state is tried, until a workable state is found (or we run out of choices).
This makes each round and its associating state's lifeline easier to reason with. Each successful forward round would create a state with a pin in the stack. A round ending with backtracking will not result in a new state in the stack, but have a previously-pinned state unpinned (and modified), with everything newer than it removed.