[PATCH] D27835: Add support to llvm::Twine for storing formatv objects

[David Blaikie via Phabricator via llvm-commits]

dblaikie added a comment.

Could you explain what makes each of the test cases interesting? (what they're testing that other test cases aren't) - it seems like a bit more testing than I'd expect, but I'm probably missing something about the mechanics of Twine.

Also - would it be worth adding a test case to demonstrate that it truly is lazy? If you introduce a type with a custom printer that has a side effect (such as modifying the object as it prints) - then create a Twine of a formatv, observe the side effect hasn't occurred, then stringify the Twine and observe that the side effect has occurred.


https://reviews.llvm.org/D27835