environ problem in 3.8p1

[Damien Miller]

Damien Miller wrote:
> Damien Miller wrote:
> 
> 
>>In his ignorance, on Thu, 11 Mar 2004, Damien Miller mistakenly wrote:
>>
>>
>>
>>>No, the protocol does not include a way to transmit more than the
>>>terminal type ($TERM).
>>
>>Markus pointed out that I am wrong: protocol 2 has a request to pass
>>environment variables, which we don't implement.
> 
> There is now a patch to implement environment passing in:
> http://bugzilla.mindrot.org/show_bug.cgi?id=815

For those who are interested, this patch has been comitted. CVS HEAD now
has the ability to pass environment variables from the client to the
server.

The variables that are sent can be controlled by a new "SendEnv" option
and the ones that the server accepts is controlled with an "AcceptEnv"
option. Multiple uses of each of these are allowed, and both options
allow basic "*" and "?" wildcards.

Be careful using these options - there exists the possibility of
unwanted exposure of information on the client side and the possibility
that pre-execution envrionment twiddling could bypass restricted
environments.

-d