[rescue] Need PCI DIO (National Instruments, Sealevel, etc) card!

[Curious George]

Hi, Bob,

> It handles some of the complexity in GPIB communications. 
> Basically, you open a tcp/ip connection (telnet works) and 
> thow commands at it.

Ah, OK.

> Some commands are to control the device itself, others are to

By "device", here, you mean the Prologix gizmo (?)

> relay the GPIB device commands to the instrument (which are 
> also pretty non-standard and instrument specific).

Yes.  But, to be fair, even as "standardized" as SCSI has
become, there are always the "unanticipated" issues that
manufacturers have to squeeze in that the Standard can't
address.  :<

The problem arises when you get folks "playing cowboy" with
their interfaces (e.g., the "AT" command sets that modem
manufacturers freely mangled to suit their interests)

> I was able to get an old Tek 7854 scope talking to a
> pc in about an hour with the Prologix adapter.  The problem
> is that my code will only work with this device and there 
> are a lot of guys on the tekscopes list with this scope
> and without the Prologix adapter.  I

Ah, understood.

> want to rewrite the code to work with a standard interface but don't
> have a card to do so.
> 
> I think it would be very interesting if somebody ever came
> out with a GPIBFS that worked similarly to the OWFS / One Wire
> File System.

Inferno tries to eliminate many of the incantations that
are tied to specific implementations out of its device interfaces.
E.g., to control a TV device, you open /dev/tvctl and throw
commands at it like:  "brightness 23", "volume 34", "init",
etc. (i.e., all of the "commands" are ASCII strings)

> The major problems are the inability to anticipate the commands 
> used by a particular instrument.

Exactly.

> I think by 488.2 standard they have to implement some common
> commands, but the command to dump the value of say the X
> register on my scope (which can be an array of data, a single 
> digit, or alpha character), will be different than the command 
> to dump the voltage info on my volt meter.

Yup.

> Also, the instruments all have different buttons that can be
> "pushed" via GPIB, each with a different name and no standard naming
> conventions that I'm aware of.  When coding for these instruments,
> there is usually a page or two (or 10 depending on the instrument)
> with tables of the commands.  Without that it would be impossible.

Yes.  The *small* advantage that Inferno's approach gives is that
the "programming" can be reduced to a bunch of text strings.

Gee, I can't wait for the "Home of the Future" (and the mess
that will accompany *that*!)