Re: Re: Re: Re: Re: [PATCH] http: avoid auth failure on a duplicated header

2014-07-18 Thread Daniel Stenberg
On Thu, 17 Jul 2014, Michael Osipov wrote: Yes it should! But you're expressing this funnily. If if _does_ probe first, it will disclose the exact same information if the server asks for basic auth Haven't noticed that I brought some fun into it. "funny" in the meaning of "strange" or "pecu

Re: Re: Re: Re: Re: [PATCH] http: avoid auth failure on a duplicated header

2014-07-17 Thread Michael Osipov
> Von: "Daniel Stenberg" > On Thu, 17 Jul 2014, Michael Osipov wrote: > >> Yes, because you're asking for it! > > > > Then I would at least require the docs to say that preempive is is > > performed > > by default. Users should be aware that they could disclose information. > > Yes it should!

Re: Re: Re: Re: [PATCH] http: avoid auth failure on a duplicated header

2014-07-17 Thread Daniel Stenberg
On Thu, 17 Jul 2014, Michael Osipov wrote: I'm fully convinved you will find servers out there returning headers like that. Maybe true but that is not covered in libcurl also. You cannot scope the auth. Right, but that's a separate limitation. It has been worked on in the past but it was n

Re: Re: Re: Re: [PATCH] http: avoid auth failure on a duplicated header

2014-07-17 Thread Michael Osipov
> Von: "Daniel Stenberg" > On Thu, 17 Jul 2014, Michael Osipov wrote: > > > WWW-Authenticate: Basic ream="A" > > WWW-Authenticate: Basic ream="B" > > > > That makes no sense and is incorrect. > > Is it really? What if it has two overlapping realms and offer you to login to > any of them to acce