Netlab device naming change around node_xx and node-xx

[snuffy22]

We have a series of checks to help us not break when using DNS etc of converting node_xx into the valid form of node-xx

Am I right saying the main reason we are using node_xx is because of our shorthand use when making lab topologies?

I know it would be a breaking change but could it be simpler to convert them to the DNS style on generation then remove some of our jinja template foo that is doing the conversion at times?

Thoughts?

[ipspace]

Am I right saying the main reason we are using node_xx is because of our shorthand use when making lab topologies?

The "original sin" was trying to use identifiers that match the usual programming language restrictions.

I know it would be a breaking change but could it be simpler to convert them to the DNS style on generation then remove some of our jinja template foo that is doing the conversion at times?

That would either break or complicate everything downstream of "netlab create". As the simplest example, "netlab connect" would either not work with the node name as specified in the lab topology, or have to track what the "original" node name was and what's actually running. There are tons of other places that allow users to specify node names.

Having said that, I'm all for removing redundancies in the templates. For example, we could use node._dns_name as the DNS-safe version of node.name. Should we start with that?

[snuffy22]

That sounds like a much better idea.