Re: aao record behaviour

[Marty Kraimer <mrk@aps.anl.gov>]

At aps we have not used the aai or aao records. Thus I fear making changes. If
You and Eric can agree on changes (starting with the 3.13.2 version) I will put
the new version in the next release.

Here is another problem for  aai and aao.

The init_record looks like

    if (pass==0){
        if(paai->nelm<=0) paai->nelm=1;
        return(0);
    }

And then it waits for device support to allocate space for the array during pass
= 1. This can lead to severe problems since storage MUST be allocated during
pass 0. If another record has a link to the array and this record gets
initialized before thye array record bad things will happen. If you and Eric
decide to fix the problem I have an idea that may work. It dose mean, however,
that associate device support also needs to be modified.

Marty Kraimer

Korhonen Timo wrote:
> 
> Hi all,
> 
> I wanted to use the aao (array analog output) record and noticed a
> problem: when I put an array to the channel that is shorter than
> the original NELM value, the NELM gets set to the number of put
> elements.
> A search in the tech-talk archive resulted in the following hit
> (from April, 1996!):
> 
> >From: "Eric Bjorklund, NPSM" <BJO@accr.lcs.lanl.gov>
> >Date: Thu, 11 Apr 1996 12:19:54 -0600 (MDT)
> >Subject: Possible Problem with aao Record
> >Message-Id: <960411121954.4c602e5e@accr.lcs.lanl.gov>
> >
> >
> > We have discovered what may be possible bug with the aao record.  The
> > problem appears to be in the "put_array_info" routine, which stores the
> > number of points to write in the NELM field of the record.  The problem
> > with this that if you write less than NELM points to the aao record,
> > any further connections to this record get the new, lower, value of NELM
> > (through "cvt_dbaddr" or "get_array_info") and are no longer able to
> > write as many points as the record originally specified.
> >
> > My questions are:
> >
> > o Shouldn't "put_array_info" be updating the NORD field instead of the
> >   NELM field? (if so, it means we should probably change the code in
> >   devAaoCamac.c as well).
> >
> > o What is the correct value, NELM (the capacity of the record) or NORD
> >   (the number of points actually written) to return from "cvt_dbaddr"
> >   and "get_array_info"?
> >
> > o (and, BTW...  shouldn't SIOL be defined as an OUTLINK on an aao
> >   record?)
> >
> >                                        Eric Bjorklund
> >                                        AOT-6 (LANSCE Operations)
> >                                        LANL (bjo@lanl.gov)
> 
> I tried the changes that Eric proposes and then the aao seemed to work
> as I expected.
> 
> Have I missed something, is there a real reason for that behaviour or
> have the fixes been left out from the distribution (I am using 3.13.2)
> by mistake?
> 
> Timo
> 
> Timo Korhonen  PSI (Paul Scherrer Institut), SLS
>                CH-5232 Villigen PSI
>                tel + 41- 56 3103262  fax + 41 - 56 310 3151
> e-mail:        timo.korhonen@psi.ch