Alan Zeino

Just a heads up, I'm testing this at the moment to see how much faster it is; we have 20k snapshot tests so we should have some decent numbers soon....

How does this compare to #140?

I don't think we ever tried this because running on device is more friction than running on Simulator. If you have some potential fixes, feel free to PR them.

I don't think any video content should be inside a snapshot, considering video compression — I can only imagine it causing you issues down the line

Likely we will at some point provide some tools for this, though the linked above project seems decent (it doesn't support this library though, only the pointfree snapshotting tool).

We have something similar inside our fork of the library because we've built the same infrastructure that Facebook has, I'll try PR a few things back here.

Could we not implement this via the `_recordSnapshotOfViewOrLayer:...` path and not the `_performPixelComparisonWithViewOrLayer:...` path? Right now, by going through the `_performPixelComparisonWithViewOrLayer:...` path, you're relying on `saveFailedReferenceImage:` to save the image...

Yep I totally want the tests to fail in auto record mode; I just wanted this implemented in a more obvious way like the other mode (‘record mode’) for code...

To answer the general question there; yes the intention of the library was to be used in conjunction with source control tools. At Facebook, they use Phabricator which allows engineers...

> @reidmain As an outside contributor I the repo will not allow me to add requested reviewers. I'm curious to know if there's anything else needed from me in order...