Re: Proposal to write a serial driver

[Scott H. <st...@ma...>]

At 22:23 +0100 1/16/05, Martin Henz wrote:
>While  most  people  are  talking about the licencing issue,
>nobody  seems  to  talk  about  disadvatages  of  the actual
>NI-VISA implementeation and the way it works.

Good point.  The VISA implementation is to keep generality and isolate what is specific to certain interfaces.  This takes careful thought.  There are some major problems I ran into with NI VISA (wait till next week when I post some timing performance issues to info-labview).

>Start  your  MS Win PC, and if the OS is up and running plug
>in a USB to serial converter, then start a labview and place
>a  VISA  control onto the front panel. This VISA control did
>not  show the new serial port. You are also unable to access
>this  serial port trough VISA. - Ok, I have tested this only
>on two different PC's, both with win XP.

Get a better OS?  To be honest, OS X is written from the ground up to handle networks going in and out and USB, etc.  I can add serial ports  on the fly to OS X.  While a VI is running and it will pick them up. The VISA control doesn't always show them, but they can be opened and used.  I try not to show my bias too often, but here, the machines are same cost, better capabilities, and a *LOT* less pain to use.

>VISA  Lock  and  Unlock are working as described, but that's
>not  a  useful  locking  mechanism for the serial port on MS
>Win.  Nobody  needs  to  give  a process exclusive access to
>anything he already has.
Well it is general and can work on other OSs usefully!.  It also has some weird effects that recently surprised me!

>The  Termination  Character  is  a  nice thing, but for some
>devices  with horrible protocols I would prefer to have more
>than   one   Termination   Character   and/or  one  or  more

The NL serial library had multiple characters, but not multiple strings.  A good thing to implement!

-Scott