Because we are using ssh multiplexing to increase ssh performance on *nix machines, we need to come up with a better way to handle ssh socket names.
Currently sockets are stored in $XGD_CACHE/nvim/netman/tmp, however this path can (after having the ssh connection information attached) result in a pathname that is too long for the OS to handle. This failure results in a silent connection failure. The end user receives no notification, it just "doesn't work".
I don't know the best way to address this off the top of my head. Currently we are using
ControlPath=$HOME/.cache/nvim/netman/tmp// %h-%p-%r.
We might be able to get away with a shorter ControlPath args, or we might be able to get away with a different storage location for these connection sockets. Maybe some logic to handle "Oh that is going to be too long, use a different hash". Not sure yet.
Because we are using ssh multiplexing to increase ssh performance on *nix machines, we need to come up with a better way to handle ssh socket names.
Currently sockets are stored in
$XGD_CACHE/nvim/netman/tmp
, however this path can (after having the ssh connection information attached) result in a pathname that is too long for the OS to handle. This failure results in a silent connection failure. The end user receives no notification, it just "doesn't work".I don't know the best way to address this off the top of my head. Currently we are using
ControlPath=$HOME/.cache/nvim/netman/tmp// %h-%p-%r
. We might be able to get away with a shorter ControlPath args, or we might be able to get away with a different storage location for these connection sockets. Maybe some logic to handle "Oh that is going to be too long, use a different hash". Not sure yet.