Re: [Help-gnucap] gnucap development snapshot 2008-02-11

[a r]

On Feb 13, 2008 4:03 AM, al davis <address@hidden> wrote:
>
> You can do:
>
> .param alpha=.1
> .op
> .param alpha=.2
> .op
> .param alpha=.3

I tested it a bit. Everything seems fine except for problems with
error detection.

When I tried to refer to an undefined parameter I got a warning:

"parameter p_trtf-WRONG has no value".

However, the simulation then continued with bogus settings. See the
attached example for details - the undefined parameter (tr time of a
"pulse" card) has been assigned values from 140ps to 250ps. Note that
the example requires a fix to the bm_pulse model that I posted here
earlier today.

I think that in case of encountering an undefined parameter Gnucap
should simply quit or refuse to run an analysis when used in
interactive mode. AFAIK all simulators treat such a situation as an
error so there is no danger of breaking compatibility with third-party
models.

Regards,
-r