We have an application where large (up to 60kBytes) blocks are transmitted
with rpc. Parts of it are arbitrary length array of chars. Now we like to
reduce the number of times where this data have to be copied. On the
consumer side we’re thinking to use shared memory. How can we force the rpc
library (e.g. any xdr_ routine), to place the incoming data directly into
this shared memory?
There is only one remote procedure with this amount of data.
What other possibilities are there? What is easiest to accomplish.
Help is greatly appreciated.