On Thu, Feb 18, 2010 at 09:52:02PM +0100, Corinna Vinschen wrote: >On Feb 18 15:39, Charles Wilson wrote: >> I ran across an interesting "feature" of remote access today. I was >> ssh'ed in to my cygwin computer, under my normal windows/cygwin account >> name, and tried to run 'patch': >> >> $ patch -p1 -R --dry-run < ../some-patch.patch >> patch: **** Can't create file >> /c/Users/CYG_SE~1/AppData/Local/Temp/poFOD7WH : Not a directory >> >> $ echo $TMP >> /c/Users/CYG_SE~1/AppData/Local/Temp >> >> $ echo $TEMP >> /c/Users/CYG_SE~1/AppData/Local/Temp >> >> $ echo $TMPDIR >> >> >> Obviously, my regular user doesn't have access to cyg_server's AppData >> directory. This is easily fixed, of course, by setting $TMP=/tmp (or >> /c/Users/<me>/AppData/Local/Temp, if you like). The question is, should >> this be something that is done by default in /etc/profile (e.g. part of >> the base-files package)? > >In contrast to other systems, sshd for Cygwin preserves a couple of >environment variables from the parent sshd process running under the >cyg_server account. The list of preserved variables is: > > ALLUSERSPROFILE > COMMONPROGRAMFILES > COMPUTERNAME > COMSPEC > CYGWIN > NUMBER_OF_PROCESSORS > OS > PATH > PATHEXT > PROCESSOR_ARCHITECTURE > PROCESSOR_IDENTIFIER > PROCESSOR_LEVEL > PROCESSOR_REVISION > PROGRAMFILES > SYSTEMDRIVE > SYSTEMROOT > TMP > TEMP > WINDIR > >Is it time to reduce this list? Should sshd remove TMP and TEMP? >Anything else?
I'd say that it should only preserve COMSPEC, CYGWIN, SYSTEMDRIVE, SYSTEMROOT, WINDIR, and, I guess, PATH. The other stuff is pretty useless anyway. You can't reliably depend on, e.g., PROCESSOR_REVISION environment variable to be accurate any more than you can rely on email that says "Certified spam free" to be spam free. I hate to see the environment polluted this way but, then, linux does it too. cgf -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
