Hi Oliver:

Am 25.01.14 22:26 schrieb(en) Oliver Eichler:
I am really amazed how everyone is so enthusiastic to provide informaton that 
is of no concern of the one requesting it. I thought after the Snowden disaster 
people start to think twice on how much details have to be supplied.

I fully agree with the approach to limit leaking unnecessary information as far 
as possible.  However, to be honest, I do not fully understand which 
security/privacy implications sending the user agent might have.  IMHO, the 
critical information in an OSM request is (a) the source IP address and (b) the 
tile id, in particular if the NSA/GCHQ/whoever links it with more meta data 
from other channels like e-mail etc. - what they probably do, as we know after 
the Summer of Snowden!

I don't see the value for OSM, though.  Or do you think they (a) store the data 
and (b) share it with the secret services?

The trivial protection is using an anonymiser proxy like TOR (with the drawback 
that it's slow, and OSM might easily blok TOR exits).  It would be great if OSM 
would offer reading data through a fully encrypted channel (i.e. https), but 
afaik this is not possible.

Maybe you give more details about your concerns regarding the user agent 
information, compared to IP address plus tile id?

Techhnically there is no reason to transmit the user-agent information at all. 
As all information is simple data, there is no need to adopt the result to the 
used user-agent. [...] The only reason to supply it is because the OSM tile 
service requires it.

This is not completely true.  RFC 2616, sect. 14.43 states that "user agents SHOULD include 
this field [User-Agent] with requests".  According to RFC 2119, "SHOULD [...] mean[s] 
that there may exist valid reasons in particular circumstances to ignore a particular item, but the 
full implications must be understood and carefully weighed before choosing a different course".

Just a side note: In the company for which I'm working, I use the user-agent 
information in the (squid) proxy as to block connections which should be 
considered as being potentially dangerous (e.g. Skype, clouds, and similar).  
Thus, this header field *does* actually have some (though really limited) value 
for improving security.

I do agree with you, that the user-agent string should be compliant to the 
specification. I will change that. But I will take my freedom to keep the agent 
anonymous by mimicking a 0815 agent.

Actually, you don't do that if you just state it's "Mozilla"!  All browsers by default 
sent *much* more information [1, 2].  Thus, a user-agent which looks so different from a 
"real" UA's identification is already a clear indication of a faked one.

Best, Albrecht.


[1] <https://panopticlick.eff.org/>
[2] 
<http://www.heise.de/security/meldung/Fingerprinting-Viele-Browser-sind-ohne-Cookies-identifizierbar-1982976.html>

Attachment: pgpqW4WCCpaFl.pgp
Description: PGP signature

------------------------------------------------------------------------------
CenturyLink Cloud: The Leader in Enterprise Cloud Services.
Learn Why More Businesses Are Choosing CenturyLink Cloud For
Critical Workloads, Development Environments & Everything In Between.
Get a Quote or Start a Free Trial Today.
http://pubads.g.doubleclick.net/gampad/clk?id=119420431&iu=/4140/ostg.clktrk
_______________________________________________
Qlandkartegt-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/qlandkartegt-users

Reply via email to