Bugs

When defining aliases like

alias pdb plot db( '!:1' - '!:2' )

you must be careful to quote the argument list substitutions in this manner. If you quote the whole argument it might not work properly.

In a user-defined function, the arguments cannot be part of a name that uses the plot.vec syntax. For example:

define check(v(1)) cos(tran1.v(1))

does not work.

If you type plot all all, or otherwise use a wildcard reference for one plot twice in a command, the effect is unpredictable.

The asciiplot command doesn't deal with log scales or the delta keywords.

The hardcopy command writes plot(5) format unless hcopydevtype is set to postscript, and plot(5) files need a program that understands that format.

BJTs can have either 3 or 4 nodes, which makes it difficult for the subcircuit expansion routines to decide what to rename. If the fourth parameter has been declared as a model name, then it is assumed that there are 3 nodes, otherwise it is considered a node. To disable this, you can set the variable "nobjthack" which forces BJTs to have 4 nodes (for the purposes of subcircuit expansion, at least).

The @name[param] notation works in commands that use the results of a run, such as print, let and meas, but not in the commands that watch a value while the simulation runs: trace, iplot and stop (for example stop when @r1[i] > 1m) report that the node does not exist.

The first line of a command file (except for the .spiceinit file) should be a comment, otherwise SPICE may create an empty circuit.

HSPICE model libraries: .params, subcircuit parameters and temper are evaluated when a circuit is loaded (and again by reset and .step), so a later let of a parameter or set temp doesn't change the circuit until it is reread. The diode's IK, IKR and TLEV, and the BSIM4 instance parameter MULU0, are not supported. .TEMP cards are ignored; use .OPTIONS TEMP=.