Giles Atkinson

Diarmuid , Thanks for running that, but I too see nothing significant, so at this point I am guessing. A few ideas: Are there enough clocks? If I interpreted it right, the tran simulation only runs for 20 cycles. Is that enough for something to happen? Also, try "eprint clk" to be sure it is working as expected. You can check the process memory to be sure the Verilator code has been loaded as expected. Run the netlist interactively, then stun the process with control-Z, run "ps" to get the process...

"script" is the Unix utility for recording a terminal session amd part of the "bsdutils" package on Debian/Ubuntu. Unlike shell redirection, it preserves the order of mixed normal and error output.

Diarmuid, I tried to reproduce, but do not have the PDK, so that failed. Please capture the output from a failing run ("script" will keep everything in order) and post the file. G

A problem with option 3 is that the rest of the netlist has to use individual nodes (name_n) so there is still an ugly disconnection. And after thinking about it some more, I think it also requires something like option 1. My current thinking is that 2 is the way to go, but it needs to extend into the vector code, so that vectors can have "[]" in their names and still be indexed. But even a prototype seems a distant prospect ... The good news is that in Xschem you can use bus nodes: they are expanded...

A problem with option 3 is that the rest of the netlist has to use individual nodes (name_n) so there is still an ugly disconnection. And after thinking about it some more, I think it also requires something like option 1. My current thinking is that 2 is the way to go, but it needs to extend into the vector code, so that vectors can have "[]" in their names and still be indexed. But even a prototype seems a distant prospect ... The good news is that in Xschem you can use bus nodes: they are expanded...

Back on topic, I think it may make sense to fix this in the controlled-source code and avoid copying a potentially-large vector on every iteration. But there may be instances of the bug in other devices.

Spaces in string parameters of XSPICE code-models

Should be fixed by a9950f0e8c.