Hi Ed & James
Consider it a single data point, but one that represents a number of tests,
including compiling with both opendap and nco, and testing nco via both local
files and opendap-mediated access.
Altho I haven't yet run it thru valgrind, it looks like it links with nco and
opendap much like the 3.6 series and seems to have the same functionality. It
does not seem to have any explosive memory leaks ;).
The netcdf4.0-linked code passes all the local file benchmarks equally as well
as the code linked to netcdf 3.6.0 and works at least as well on the
OpenDAP-retrieved data. BY that, I'm testing it vs a system equally remote
to the DAP server (the server is linked against the netcdf 3.6 lib), but with
code built with the netcdf 3.6 libs.
Neither works perfectly (seemingly due to latency and timeout issues - still
have to debug this. If the nco benchmark code is run on the same machine as
the server, it runs to completion just fine, if slower than accessing the
local files, as you might guess), but the netcdf4.0-linked code gets further
thru the benchmarks.
FYI...
--
Cheers, Harry
Harry J Mangalam - 949 856 2847 (vox; email for fax) - hjm@xxxxxxxxx
<<plain text preferred>>
