This results in the icon fetch failing. If you scan and fetch a second time on that redirected path, you can see that the new domain does have a valid icon and works.
This actually looks like a bug or a regression in DownloadImages, so it should be fairly simple to fix.
Ah - whoops. I was looking at the wrong code. This was an issue with the new HttpClient branch, not the mainline. Will include fix as part of the async/httpclient pull request
When scanned, some domains return what appears to be valid icon urls. For example, a scan of Microsoft's live.com returns two URIs:
But attempting to download images at those URLs redirects to a new domain, which resolves to a text/html document:
This results in the icon fetch failing. If you scan and fetch a second time on that redirected path, you can see that the new domain does have a valid icon and works.
This actually looks like a bug or a regression in DownloadImages, so it should be fairly simple to fix.