Redirect handling is now transparent to clients of fetchcache.
The new scheme works as follows:
1) Request content for URL (fetchcache()
2) Start fetch of content (fetchcache_go()
3) If no redirect, continue through LOADING, READY, DONE etc. states as before
If redirect, receive NEWPTR for each redirect that occurs, then continue
through LOADING, READY, DONE etc. states as before.
The upshot of this is that redirects result in extra contents being created. It also means that, until LOADING has been received, the content (and thus the URL being fetched) may change. Therefore, fetchcache clients should expect to have to deal with transient data prior to LOADING occurring.
As a necessary side-effect of this, the HTML object URLs and CSS @import URLs are no longer stored alongside the relevant contents. These URLs can be accessed by interrogating the url member of struct content anyway, so it was a rather redundant scheme before.
svn path=/trunk/netsurf/; revision=3787
78d194cb77