Fix flatten optional assignment #142
Merged
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
The single-pass version of flatten_adaptor currently does a move-assignment of an optional, which doesn't work if the inner sequence is not move-assignable -- for example if it involves a capturing lambda. This is rare, but I ran into it today doing Advent of code...
For copy/move assignment between two engaged optionals, std::optional always use's T's assignment operator. For flux::optional, if T is not copy/move assignable, then we fall back to destroying the contained LHS and copy/move constructing from the RHS. This was part of my justification for providing optional references back in the Flow days, and it's still appropriate today.
We've had optional<T>::emplace for a while. As you might expect, it constructs a T directly inside the optional, using the supplied arguments. For the optional<T&> version, we require the argument to be a single argument of a "compatible" lvalue reference type -- that is, the same condition as the constructor. This allows emplace to be used in generic code which doesn't know if it's dealing with optional<T> or <T&>.
Rather than constructing a RHS optional and using assignment, we'll use emplace() to construct the object directly in the LHS, which results in fewer operations. It also means the changes to the optional assignement operators we're necessary, but never mind...
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #142 +/- ##
=======================================
Coverage 97.98% 97.98%
=======================================
Files 66 66
Lines 2380 2383 +3
=======================================
+ Hits 2332 2335 +3
Misses 48 48 ☔ View full report in Codecov by Sentry. |
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.
The single-pass version of
flatten_adaptorcurrently does a move-assignment of an optional, which doesn't work if the inner sequence is not move-assignable -- for example if it involves a capturing lambda.This is rare, but I ran into it today doing Advent of code...
This PR fixes it in two ways. Firstly, in our
optional<T>implementation, ifTis copy-constructible but not copy-assignable, we destroy + construct in the assignment operator (and likewise for move). This allowsoptional<T>to be copy/move-assignable even whenTis not, for example if it's a capturing lambda.Secondly, we actually remove the problem by using
optional::emplaceinflatten_adaptor, which results in fewer total operations anyway, but does require addingoptional<T&>::emplaceso things don't break.