[Vimprobable-users] Download bugs

[Morgan H. <mt...@gm...>]

jasonwryan came across a regression in download handling, where 'f'ing
or ';s'ing a pdf link does nothing. After a bisect, I found this was
introduced by:

commit cbf47b201a04aaf5165595c308dd76bb42b185a9
Author: Hannes Schueller <ha...@yl...>
Date:   Fri Dec 26 11:19:24 2014 +0100

    bugfix: only trigger a download when status code is in the 2xx range

What I found is that SOUP_STATUS_IS_SUCCESSFUL always fails because
the http_status is always set to the default of 0. Some cursory
searching about this leads me to believe we should also be checking
SOUP_STATUS_IS_REDIRECT, but that obviously doesn't matter either
until the primary bug is fixed.

While bisecting this I also came across another bug in downloading
which does not appear to be new. When attempting to download some of
the direct pdf links from google (for an easy example) you'll get ugly
urls like "www.google.com/url?target=http://www.example.com/file.pdf".
webkit_download_get_suggested_name gives us a suggested filename of
"url", and the download fails. This seems more like a bug in webkit,
or possibly in our usage of the webkit interface.

Morgan