On Thu, Jul 07, 2005 at 05:08:54PM +0200, Robert Lemmen wrote: > hi everyone, > > i have another approach on this one: > 1. check the content type of the download, call a special handler if it > is text/html (will also solve #303958)
I agree that #303958 is important and should be fixed, it is really annoying. However I would emit an error message if the final file (i.e. after redirections) to download is text/html (or better check with file(1) or examine the first N bytes). > 2. this handler should check the host part of the download url and > switch to a site-specific handler > 3. the sf handler would then parse the html file, extract the mirrors > and retry from one of them I like this approach, but IMO it is not uscan's work to workaround broken mirror schemes and/or watch files. This can be also considered as a separate wishlist bug for uscan, feel free to re-file it as such :) regards, filippo
signature.asc
Description: Digital signature
