This tests triggers the creation of a very large image surface. This fails because of checks, but the codepaths this triggers then continue to use the (invalid) image surface. This triggers a couple of assertions, but the NULL surface ends up being silently ignored by Cairo. This doesn't seem like it's something we should rely on though, it seems to be a result of the 'RasterImage' believing it's ridiculously big (40K+).
The GIF actually seems to have a frame that big (49173x49173), so the fundamental problem seems to be more about how RasterImage deals with the fact it is not allowed to create a new image surface of that size.
Created attachment 575423 [details] [diff] [review] Check CairoStatus of surfaces in DrawImage