[Gnucap-devel] pulse again

[Felix Salfelder]

Hi Al.

bm_pulse.{1,2,3}.ckt contains

Y1   2  3  pulse iv=20u pv=0 delay=500n period=1u
Y2   3  0  pulse iv=0 pv=20u delay=500n period=1u

but, this doesn't seem to work as intended. the period argument is
"ignored".

this apparently goes back to a fix in bm_pulse.cc, commit 8d51d947799.

- time = fmod(time,_period);
+ //time = fmod(time,_period);
+ if (time > _delay) {
+       time = fmod(time - _delay, _period) + _delay;
+ }

the rebased fmod looks much better, but changes the interpretation
slightly (intended?). now the Y's are missing a width=500n.

maybe it even makes sense to have

if (_period.has_hard_value() &&
                _delay.has_hard_value() &&
                !_width.has_hard_value()) { untested();
        _width = _period - _delay - _fall - _rise;
}

somewhere during precalc...

regards
felix