mirror of
https://github.com/libretro/RetroArch.git
synced 2026-11-12 14:36:41 -10:00
A download from these servers was gathered whole in memory, up to the 256 MB body limit, then written to the local file in one piece; the WebDAV driver carried a TODO asking for exactly this. They now use task_push_http_download_file_with_headers: the body goes to the file in pieces as it arrives, and the callback opens the finished file to hand on. The HTTP error text these drivers log still reaches them. A body with no Content-Length or chunked framing is still refused, and its file is removed with it. A 404 (or S3's 400) still means the server does not have the file; the download removed the file already.