Open jleibs opened 3 months ago
On a related note, I've wanted to be able to use the CLI as a data forwarder tool for a long time, e.g.: rerun my_data.rrd --connect
(or, in this instance, rerun my_blueprint.rbl --connect
):
The easy solution is to always spawn a TCP server when we launch the rerun
binary (unless --no-server
or whatever),
but we must handle the case of an existing Rerun already running, hogging that port.
We can either:
0
port (ask the OS to pick a free one)debug
(don't bother the user too much)
We don't appear to start the TCPServer if we pass a local file to rerun on start.
Although we've always had this issue when loading
.rrd
files from the CLI, the fact that Blueprint files and RRD files combine makes this particularly annoying.Consider the following workflow:
I've saved a blueprint.
Now I want to run:
to restore it (related issue: https://github.com/rerun-io/rerun/issues/5694)
However, because there is no TCPServer running, I can't run an app that connects to it.