You can't make that work generally I think. You'd have to have another region of shared memory established. That feels like it should be a totally different transport. Indeed with shared memory you would really like to establish a protocol for placing data in the shared memory regions. Anyway nanomsg doesn't have anything like this at present. Sent from my iPhone > On Apr 15, 2014, at 1:31 AM, Martin Sustrik <sustrik@xxxxxxxxxx> wrote: > > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Hi Garrett, > > >> What was the purpose of the extra byte in the IPC header that >> precedes the length (for each message). If not for that one >> difference, the message formats between IPC and TCP/TLS would be >> identical. I’m having a hard time seeing value in that extra >> byte. >> >> (I noticed this tonight when I was trying to use IPC between C and >> my Go version, and I discovered that while I tested TCP, apparently >> I never tested IPC interop… Oops!) > > The thing is that with IPC you can actually do zero-copy (store the > data in shared memory, pass the pointer to the other process). Thus, > we need a place to signal that the message is not a standard message > passed via IPC pipe, but rather a descriptor of a block in shared memory. > > Martin > > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.11 (GNU/Linux) > Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ > > iQEcBAEBAgAGBQJTTO56AAoJENTpVjxCNN9YmOAH/i6UymZhcIh5/3C7i4QRK3kQ > DTBniByOUQosYXTrPRTf+kO/qtRenQcX5uoYQXIN+n0P74eqXsLodyPA5F6RDCRI > Ss5hPNrXw05DfWo6OMoldyEmBRrpvAB9bmhRMaMlUXrl4NvVre3Qk8Pmy3nLX0Gc > p6IOZtah/qAbLDYGZTUO3KyhIcV7Frj+Xr5Xw6MxzuFcpX8faBJ7shyOj9RdQhIR > nA9RTGRRXLmWWCZPgUm3y7ePYyCi+YhBHc26rJfDv2yF9th7NoM4j5+93AHzYkcB > 2gLj3tT/vk7zlWDPFGp+FdQyjBGVwEHRzabo4G3uQYhvdGt38ghkcqdP/YDzLAk= > =cc/i > -----END PGP SIGNATURE----- >