EdJoPaTo

> I think this likely points to the idea that assert_eq should just be a method in the buffer module. That wouldn’t change much. It would be `Buffer::assert_eq` vs the...

> Just a reminder to ground this idea, the reason `assert_buffer_eq` exists is purely because the diff between buffers is difficult to visually inspect when there's an error, and the...

funny realisation: `self.assert_lines(lines)` is even shorter and would require bigger example buffers to ensure the formatting in the sources stays similar to rendering…

Thought about adding `Buffer::assert_eq(&self, other: AsRef)` but that only removes one & per line. its not that significant and it results in less clear code that the input is only...

> When writing test code that interacts with the buffer (widgets etc.), my usual (Test after Dev) flow is: > > 1. write a broken test that asserts that the...

> I think this is ready to merge, but I'm 50/50 on whether to wait til after 0.26.3 to merge (so we can get out the unicode line bug fix)....

Any improvements for the first message (above the line) as a commit message? When not, this should be merged.

Ha, great, GitHub applies some of my comments to both answers to @joshka and as new review comments… 😒

> 1. Removing `Default` is a breaking change (which I'd be down to push), and thus requires proper deprecation Yes, but we can not mark them as deprecated directly. That's...

> * partial construction of types is something that the builder approach makes possible, so initially useless values is often an intentionally allowed state. Normally with the builder pattern there...