Re: New gnucap development snapshot
al davis <[email protected]>
| Newsgroups | gmane.comp.gnu.gnucap.devel |
|---|---|
| Message-ID | <[email protected]> |
On Wednesday 06 December 2006 23:46, Dan McMahill wrote: > Also, I need to figure out what the right thing is for > -rdynamic with that compiler. But since we don't get that > far it doesn't matter yet. You may not need it. It may be on by default. I am sure that if it is needed, something like it is provided. What it does is to export symbols to dynamic libraries loaded by dlopen. If you don't do it, attaching a model with dlopen (attach command) will fail. The way you find out if itis needed is as follows: Remove a file that could be a plugin from the makefile. Let's pick a simple one, how about the switch .. Remove d_switch.cc from the Makefile. Relink. The link should succeed, but the switch component is no longer supported. If you try to access it you will get an error message. Maybe an easier one to leave out might be "bm_pwl.cc" .. That is the PWL source that Spice has had forever. The same ideas apply. If you leave out the resistor, you might find that something else depends on it, so the resistor probably isn't a good choice. Compile that file into a .so file ... With gcc the command would be "g++ -fPIC -shared d_switch.cc". Since I didn't name the output file, it will be called "a.out". Now run the simulator. Give the command "attach a.out". The switch should work now. If something is wrong, the attach will fail, hopefully with a useful message. For another compiler, the procedure for compiling shared objects also needs to be determined. This will become an important feature in the future. For systems that can't deal with dynamic libraries, I suppose all the models could be compiled in as they are now.