On Friday 18 August 2006 17:12, Stuart Brorson wrote: > Well, I think that if your extensions are a superset of SPICE > (and not a change), then leave them in. You never know how > many users are reliant upon your extensions until you remove > them. Then you'll have crowds of angry users e-mailing you > about their missing extensions. OTOH, it doesn't hurt > anything to leave them in.
The first issue with this is that the extensions are not quite a true superset, because SPICE has a certain behavior for non-conforming syntax that is not documented. People use it, usually unintentionally. Then when an extension is there, the nonstandard usage that used to work on one SPICE triggers an extension in another, leading to incompatible behavior. The second issue is maintenance. Any code that remains must be maintained. It remains in the regressions. Sometimes there are changes that have a global effect. The code for the old extension needs to be patched, tested, etc. > As for gnucap's Verilog-AMS ability, I was wondering how you > plan on handling the ability to read/handle the new file > format? Can you mix Verilog-AMS and SPICE together in the > same file? Or are Verilog-AMS and SPICE files incompatible? They are completely different, but mostly non-clashing. You will be able to mix in the same file, but only on a block basis. For example, if a block begins with ".subckt", it is SPICE syntax until it sees ".ends". Likewise, if a block begins with "module", it is in Verilog syntax until it sees "endmodule". Certainly, some issues will show after it is out, such as beginning a SPICE netlist with a MOSFET named "module". _______________________________________________ Gnucap-devel mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnucap-devel
