Re: lexical analyzer for .db?

[Ralph Lange <Ralph.Lange@bessy.de>]

Hello Heinrich,

If you additionally set the PHAS field of all input records to a low 
number (e.g. 1) and the PHAS field of all calc records to a higher 
number (e.g. 2), you will ensure that the .2sec-task will first process 
all inputs and then process all calcs.

That way your calcs will always see all input values from the same run 
of the .2sec-task, so they will always be newer than (0.2 seconds plus 
run time of .2-sec-processing).

Possible exceptions:
Asynchronously processing inputs: calcs see the old value.
Overloading the 0.2sec-task (processing all .2sec- records takes longer 
than .2 sec): major screw up.

Difference to FLNKs: FLNKs ensure that each pointed-to calc record is 
processed directly and immediately after the FLNKing input record, 
setting the PHAS field to (1,2) only ensures that all calc records (in 
any order) are processed after all the input records (in any order).

Ralph


Heinrich du Toit wrote:
> Hi
>
> Thanks for all your help
>
> Wonder why the FLNK isn't a direct DB link then if possible? (e.g. 
> record inside same IOC)
> Does the fanout do it differently?
>
> What would you say about the idea of just putting all the inputs on 
> SCAN = .2 second
> and then all the calc(out) rule records also on .2seconds SCAN ?
>
> The way I see it a calc record can then never work with a value older 
> than .4 seconds... which is fine for the system I think.
>
> Can maybe just FLNK or fanout the critical stuff then
>
> -H