Re: [Bug-wget] Enqueue logic problems

[Tim Rühsen]

Am Donnerstag, 2. Mai 2013 schrieb Micah Cowan:
> > On Thu, May 2, 2013 at 9:00 PM, Tim Ruehsen <address@hidden> wrote:
> >>
> >> Darshit, I guess you are talking about redirection.
> >>
> >> That is 'wget -r gnu.org' is being redirected to www.gnu.org (via 
Location
> >> header). Wget now follows the redirection, but only downloads index.html
> >> since
> >> all included URLs in index.html refer to www.gnu.org. But we requested
> >> stuff
> >> from gnu.org.
> 
> Ah, yeah that's a decent point. I like it, but then, we run into
> name-trusting problems along the lines of why --trust-server-names was
> introduced, if we just happily translate a hostname to its redirection
> (and in particular, begin accepting more pages with that unasked-for
> hostname). Obviously, using --trust-server-names solves the issue, but
> I'm not sure requiring that is any better, from a user experience,
> than adding -D -H.

Right now, using --trust-server-names does not solve this issue (it just makes 
wget trust the given filenames).
But we could use it to solve it.

[BTW, -D doesn't solve the issue either. Try 'wget -d -r -D www.gnu.org 
http://gnu.org. Not shure if -H -D www.gnu.org only spans to www.gnu.org and 
not to any other domain.]

Specifying --trust-server-names .wgetrc let it work for each invocation of 
wget.
Having -H (without -D) in .wgetrc is too dangerous...
Having -H (with -D) in .wgetrc is too specific. The user must change .wgetrc 
for every use case with different domains.

Using --trust-server-names / --no-trust-server-names for redirection domains 
would be nice.

Tim