Re: WHAT IT MEANS
There is no hope for a netlist standard for spice because each vendor add there own features to follow the user requirements. Examples are: parametric subcircuits, namespace conventions, simulation options, result storage options and uncounted model flavors.
There is an actual requirement for a standard which is driven by automatic characterisation and exploration tools. In this usage spice is only used as an engine driven by a higher layer design enviroment. This enviroment could run for example 2 HSIM for functional, 8 Spectre for DC, 6 ADS for HB and 46 Spice3F4 on a cluster system for a complex mixed signal circuit verfication. The interface is then a specific ripper for the commercial engine. It drives the netlist, commands and collects the results.
The next level of interface bypass the spice netlist and command reader and use a memory API to drive direct the matrix engine. I expect that this come only with a single vendor because there is a dramatic drop of understanding of the advantages of this approach in the design community. Furtheron the single big company trend defocus interop standards.
I have heard that there is a TCL-Spice, so a spice driven by a TCL-Script interpreter. But I think it is more focused on automation features.