I think there are two possibilities here - either the starter script
gpsprune doesn't parse the proxy information properly and passes the
wrong information to GpsPrune, or the networking stuff used by GpsPrune
doesn't take any notice of what's passed in. It would be very useful to
know whether starting GpsPrune from the command line with
java -Dhttp.proxyHost=10.0.0.1 -Dhttp.proxyPort=3128 -jar
makes things behave as you want.
Looking at Debian's gpsprune script (in 10.1), I can't see any reason
why it wouldn't pass the port properly through - if I give it
http://10.0.0.1:3128 then it appears to correctly split up the host and
port, and pass both successfully onto GpsPrune. So knowing whether the
direct specification works or not would help rule out the gpsprune
script as the problem.
Another problem I see in this script (although I'm not too familiar
with bash), is what happens when you give a host but no port? In the
case of "http://10.0.0.1", it correctly identifies the host as
"10.0.0.1", but the second sed command fails to find a match, and
returns the whole $http_proxy variable. So I end up with host=10.0.0.1
and port=http://10.0.0.1 - surely that can't be right. Perhaps it is
rejected later and reset to 80 by the networking classes? But I assume
that rather than checking [ -z $proxyport ], it should be running a grep
and looking at the return code to see whether a match was found or not.
Any Debian bash experts feel like commenting on this one?
Pkg-grass-devel mailing list