$PH/COOL-LCP/pr_set_keyNotes041207.txt   -  RJL041207
and $PH/04s522/lcp04s/ajlopez/lcp/olc/AISetStateNotes041207.txt
(linked to $CASE/04s522/ajlopez/lcp/olc/AISetStateNotes.txt)
NOTE: This is not an earlier version of AISetStateNotes.txt.

$CASE/gen/ver_12/sjaganat/srcMain/gen_defines.c currently has this invalid  
definition of pr_set_key as a simple assignment: [(tbl)->fld = (value); \\\n");]
(preceded by conditional logging of this change to the database).

Obviously this needs  pr_unlink* and pr_link*  calls to be added 
to generate a valid macro pr_set_key  operation on an fkey field.

Adding these calls, and testing them on e.g. Hominid or setGame 
is the project I  suggested earlier to Marina.a
For this case, the call to be tested could become
pr_set_key(AIcurr, STid,  TRcurr->STid2);
iin the lcp/olc/doTransition functionof processoneevent.c.
---------------------------
    fprintf(schh_fp,"#define pr_set_key(tbl,fld,value) \\\n");
    fprintf(schh_fp," do{\\\n");
    fprintf(schh_fp,"    if((tbl)->RFLAG==1 && /*!PARSE && */ hcg_log ==1 )\\\n"); /* Sathya */
    fprintf(schh_fp,"    {\\\n");
    fprintf(schh_fp,"       hcg_key key_value = ( value ); \\\n"); /* Sathya */
    fprintf(schh_fp,"       fprintf( hcg_logfileptr, \"PR_SET_KEY %%s %%s %%s\\n\",\\\n");
    fprintf(schh_fp,"                decode_retstr( (hcg_key*)(tbl) ),\"fld\", decode_retstr(&key_value) );\\
\n");/* Sathya */
    fprintf(schh_fp,"    }\\\n");
    fprintf(schh_fp,"    (tbl)->fld = (value); \\\n");
    fprintf(schh_fp," }while(0)\n\n");
---------------------------
However, this is not the case in  any  project olc subdirectory I know of.

The state field  of the current  Active Instance is updated 
in file $CASE/04s522/ajlopez/lcp/olc/processevent.c
by this function call sequence: 
	ProcessOneEvent(EIid) calls DoTransition(TRcurr->STid2, AIid, EIid);
	DoTransition calls ActiveInstanceSetState(AIid, StateGetName(STid));
	ActiveInstanceSetState (at activeinstance.h:43)

Any schema.sch field can be updated  by functions which are compile-time inline 
expansions of the SETROUTINE macro, defined for each updatable field of each 
schema table and field type in olc3common.h (e.g. ActiveInstanceSetState):

There are 44 refs that invoke  or comment on the GETROUTINE  macro
and 20 refs to the COPYROUTINE macro. However,  there only  6 refs to 
SETROUTINE in $CASE/04s522/ajlopez/lcp/*/*{.c,.h}: two  macro  definitions 
(but no calls) in lcp/Team1 and lcp/olc versions of olc3common.h.

------------
In olc3common.h, these macros are well-documented: 
For int and float and char* fields, SETROUTINE is used:
----------------------
#define SETROUTINE(field_type, object_name, object_abbrev, field, value_type) \
FUNCTION2(                                                              \
void P3(object_name,Set,field),                                         \
    hcg_key,      P(object_abbrev,id),                                  \
    value_type, value                                                   \
)                                                                       \
{                                                                       \
    pr_find(object_abbrev, P(object_abbrev,id), P(object_abbrev,id));   \
    P(object_abbrev,curr)->field = (field_type)value;                   \
}
-----------------------
The standard routines at  activeinstance.c:70 deliver string values
	GETROUTINE(const char *,ActiveInstance,AI,State)
	COPYROUTINE(ActiveInstance,AI,State)
However, I can find no invocation of 
	SETROUTINE(hcg_key,ActiveInstance,AI,State, char*);
The function ActiveInstanceSetState seems to have been  
custom-written at activeinstance.c:73 to do a strncpy:
-----------	
FUNCTION2(
void ActiveInstanceSetState,
    hcg_key,      AIid,
    const char *, NewState
)
{
    pr_find(AI, AIid, AIid); //pr_find updates AIcurr - RJL041206
    if (AIcurr != 0) {
        (void)strncpy(AIcurr->State, NewState, StatesNameMaxLen);
    }
}
--------------
Note added to olc3common.h: 
-------------------
/* Note by RJL 041207: TBD: olc's set macros assume strings are represented 	*/
/* in type (struct XX) by a char* heap address, not an imbedded charstring.	*/
/* This is counter-intuitive to GEN's external ASCI format is a string		*/
/* of pre-defined length (Exception: last field of type tnn may be truncated.)	*/
/* TBChecked Assumption: chgen embeds space-filled char arrays in table rows,	*/
/* so ActiveInstanceSetState was custom-coded above.			*/
/* There are more efficient ways to do this. (TBD:  a new macro  AISetState 	*/
/*  with pointer  AIcurr instead of pkey AIid as arg1.)				*/
------------------