Experimental Physics and
Industrial Control System

[email protected] (John R. Winans) · Wed, 16 Oct 1996 15:28:31 -0500

Chip Said:
>
> Jim said:
> 
> >I don't think records should have these fields.
> > One solution is to use the parm
> >field of the input link,  another solution is to use a "Config"
> >function in the vxWorks startup.cmd file (many Argonne device
> >support modules use this method).
> 
> I agree.  When we did CAMAC device support we did it generically, and
> placed this type of configuration information in the link field after
> the "@" (the parm field).
> 
> I don't think these new fields are an optimal solution, and would
> not look forward to bigger records.  I'm actually not overly fond
> of the current set of config parameters (ESLO, etc.).  Gabor's 
> point of separating device dependent from device independent --
> 
> >this description has (and need) two parts , the general one
> >and the record type dependent one. Why I keep them in one field ?
>
> -- is well taken, but I think that there need be only a total of
> 2 config fields, defining a slope and an offset, not the currently
> defined set of 4 (EGUF,EGUL,ROFF,ESLO):
> 
> 	VAL = RVAL * ESLOPE + EOFFSET
> 
> where device support must return in RVAL a properly manipulated signed
> integer, perhaps doing some integer manipulations inside the black
> box to return an unscaled but decoded integer.  Whatever constants
> device support needs to accomplish this should be put into the @parm
> field of the link.

Be careful where you are going with this... before getting off track remember
this thread started as a request for a place to same what appear to be 
hard-limits for what I assume is some type of device.  

If there is no specific device support (as in Tim Mooney's serial example) 
then the record's link field(s) is(are) the only suitable place for the 
information to go.  But if there *IS* device support written, I would 
encourage the code and config for these attributes to go there (somehow.)

One VERY good reason that I can see for keeping it out of the record is 
that it will require addressing the (new?) alarm conditions that will be 
required by these types of fields.

Before implenting anything on this topic, I urge all parties concerned to
review the purpose of the record and device support modules.  Easy to say from
here eh? ;-)  And then reapproach it.  Jim, Marty, Jeff and I have discussed
this stuff to death (and then some.)  Perhaps these newcomers may have some
fresh insight that will start the HCT ball rolling again.  Without it we (you)
may risk ending up with a 1/2-baked Rx.

--John Winans

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System