Re: subRecord fix

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

Andrew Johnson wrote:

> > It's just that it doesn't seem to make much sense to pass the
> > 'process' pointer out there - or do I miss something essential here? 
>
>
> I believe it's there for subroutines that want to cause asynchronous 
> processing - if the user subroutine sets psub->pact true before 
> returning because of a background I/O operation, it will need to call 
> the record's process() routine again when that I/O operation is 
> complete.  Passing it into the init subroutine is simpler for the user 
> than requiring them to look up the process routine than using a 
> construct that may at one time have been
>
>   (((struct rset*)psub->rset)->process)(psub);
>
> which is a bit ugly and complicated if someone is new to C programming. 
> The cast is no longer needed, but it might have been back in the old 
> GTACS days.
>
> - Andrew

The dbCommon field rset did not initially exist. That is probably why the second 
argument is passed to the inam routine. The entire code for calling the iname 
routine looks strange.

    if(strlen(psub->inam)!=0) {
        /* convert the initialization subroutine name  */
        psub->sadr = (void *)registryFunctionFind(psub->inam);
        if(psub->sadr==0) {
            recGblRecordError(S_db_BadSub,(void *)psub,"recSub(init_record)");
            return(S_db_BadSub);
        }

        /* invoke the initialization subroutine */
        psubroutine = (SUBFUNCPTR)(psub->sadr);
        status = (*psubroutine)(psub,process);
    }


Note that sadr is later used to store the address of the snam routine. I think 
the above code should be:

    if(strlen(psub->inam)!=0) {
        /* convert the initialization subroutine name  */
        psubroutine  = (void *)registryFunctionFind(psub->inam);
        if(psubroutine ==0) {
            recGblRecordError(S_db_BadSub,(void *)psub,"recSub(init_record)");
            return(S_db_BadSub);
        }
        /* invoke the initialization subroutine */
        status = (*psubroutine )(psub);
    }

Note that record reference manual never mentions a second argument to the init 
routine.