Conference iosg::all-in-1_v30

    Hi,
    
    If oa$scroll_key is blank then it is trying to tell you that the record
    being pointed to is not available. Is it just CAS that is a problem for
    you? If so try typing <debug on and following the actual logic used.
    One possibility is that the CAS code you expect to run is not in fact
    being run.
    
    Cheers
    
    Stuart
    
    p.s. If you're on V3.0 DEBUG is available to programmers, but with V2.4
    you need one of K60n patches and you need to follow the special
    instructions to get DEBUG to work.

    Stuart,
    
    Thanks for your response.  It turns out that CAS isn't the only
    problem.  After much fussing around, it seems that if I read one item
    in the index while the first item on the whole index is displayed, I
    can no longer read any other item.  When I point to another item on the
    index by virtue of the up/down arrows or by entering the item number,
    the first item I read is displayed.
    
    If I scroll down or NEXT SCREEN such that the first item in the index
    is no longer displayed, I can now read the current item in the index
    and can keep moving the pointer and reading whatever items I like.  If
    I scroll back up, or GOLD T or whatever, the next item I read will be
    the only item I can read until the first item in the index is scrolled
    off.
    
    I know this probably makes no sense to anyone (it makes no sense to me)
    but this is the observed behavior.
    
    I am on 2.4 and do not have the patch you mention.  I have turned trace
    on and know that the CAS code (may not be the problem after all) is on
    the MORE$SCROLL$KEYS$INDEX form.
    
    Joe

    Hi,
    
    What is the data set being bound to by the phantom? Is your Read code
    accessing the data set by the phantom or using the underlying data set?
    If the latter then is it possible that the underlying data set can be
    opening and closing.
    
    Perhaps it would be a good idea if you post extracts from the Named
    Data showing how you do the BIND, and any pre-processing. 
    
    I also recommend that you get the patch with DEBUG installed. It makes
    program development n times easier. You will very quickly recoup the
    time it takes to install it.
    
    Cheers
    
    Stuart

Stuart,
    
    Here are pieces of my code.  Thanks for any help you can provide.  I
    will not be installing the patch because we will be upgrading to 3.0 in
    a few weeks.  I'm trying the fix this anomaly before the upgrade.
    
    Thanks,
    
    Joe
    
    
Named data from index form:

;;~~DISPLAY_FORM~~;;

FORM LANL_OD_INDEX_FIND\
IFEXIT\
CLOSE_PRIOR\
XOP "~~BIND~~"\
GET OA$STATUS=#STATUS\
.FX IFSTATUS\\OA$FLD_STAY\
IFNOTSTATUS\
GET OA$DISPLAY=OA$_GBL_NO_DOCS\
REPEAT

;;~~BIND~~;;

.IF #ODIPH == "*ODI1" THEN GET #NEWOPH = "*ODI2"  ELSE GET #NEWOPH = "*ODI1"\
.IF #SEARCH_OD_FOLDER = "[" THEN GET #KOR=":DOCUMENT" ELSE GET #KOR=""\
GET OA$FUNCTION =
 "BIND/SFILE='FILENAME,OR,#SEARCH_OD_PHRASE' " #NEWOPH " TO CAB$" #KOR 
 " WITH .FOLDER = #SEARCH_OD_FOLDER
 AND  .TITLE = #SEARCH_OD_TITLE
 AND (.DOCNUM EN #SEARCH_OD_DOCNUM2 OR .DOCNUM LE #SEARCH_OD_DOCNUM2)
 AND  .DOCNUM GE #SEARCH_OD_DOCNUM1
 AND  .KEYWORDS <=> #SEARCH_OD_KEYWORDS
 AND (.CREATED_NBS ENS #SEARCH_OD_DATE2
  OR  .CREATED_NBS:8 LES #SEARCH_OD_DATE2:8 )
 AND  .CREATED_NBS:8 GES #SEARCH_OD_DATE1:8"\
.IF #SEARCH_OD_FOLDER = "[" THEN GET OA$FUNCTION = "FOR " #NEWOPH " DO OA$NULL"\
GET #STATUS=0\
GET OA$FUNCTION="FOR FIRST " #NEWOPH " DO GET #STATUS=1"\
GET OA$STATUS=#STATUS\
.FX IFNOTSTATUS\\GET OA$FUNCTION = "BIND_BREAK " #NEWOPH\\GET OA$STATUS=0\
IFSTATUS\
OA$SCL_EXIT\
GET OA$FUNCTION="BIND_BREAK " #ODIPH\
GET #ODIPH = #NEWOPH\
GET #STATUS=1

;;OA$_MO_READ;;

.FX CAB CURRENT OA$SCROLL_KEY\\IFNOTSTATUS\\OA$SCL_REFRESH\
IFSTATUS\
DO LANL_OD_READ\
OA$SCL_REFRESH

Read script:

! LANL_OD_READ.SCP
!
! Script to list current Organizational report.
!
! Determine type of formatting for this document

    .if OA$DIR:"*.*".%WHOLE[OA$CURDOC_FILENAME] eqs "" then .goto NON_EXIST_FILE

    .if $REMOVE_BLANKS eqs OA$Y -
	then LIST/OVERLAY=LIST_Status/PRE_FUN="FORCE"/DATA=OA$CURDOC <FORMAT -
		"@OA$CURDOC","","VT100",-2,1,1,1 -
	else LIST/OVERLAY=LIST_Status/PRE_FUN="FORCE"/DATA=OA$CURDOC <FORMAT -
		"@OA$CURDOC","","VT100",-2,1,1,0

    .if #CHOICE eqs "0" or #CHOICE eqs "-1" then .goto FORMAT_ERROR
    .exit

    Hi,
    
    I couldn't see anything wrong with that. Two more things to check:
    
    A) What about the Pre-Porcessing Named Data of the CHOICE field. This
       will be invoked after every option.
    
    B) Type <BIND_STATUS at various points. This will give a list of
       Phantoms open. Sometimes if the logic is not right you can open
       many phantoms with the same name.
    
    Cheers
    
    Stuart

    Stuart,
    
    The <BIND_STATUS showed no strange behavior.  Here is the named data
    for the CHOICE field.
    
    ;;CHOICE;;
    
    /PRE='GET SELECTIONS=OA$SCROLL_SELECTED\GET UNREAD=OA$MAIL_COUNT'
    /HARD=OA$_WP$INDEX_OPTIONS
    
    Another observed behavior.  I created a list of 14 items.  After I read
    one item (any of the 14) when I entered the number of another item or
    used the UP/DOWN keys and attempted to read another item, the first
    item I read was displayed.
    
    If I then pressed GOLD T or GOLD B, I could then enter the number of
    another item or used the UP/DOWN keys and read another item.  Until I
    did another GOLD T/B, that was the only item I could read
    (OA$SCROLL_KEY has no value).
    
    Thanks for any help,
    
    Joe

    I'm starting to run out of ideas now! Are you by any chance building an
    index of shared folders? If this is the case there are some bugs with
    doing this that are fixed in K603.
    
    Cheers
    
    Stuart

    Stuart,
    
    By jove, I think you've got it.  The documents being indexed are in a
    shared folder (2.x style).  You say these bugs are fixed in K603, can I
    assume they are also fixed in 3.0?  If so, I'll let this problem ride
    until the upgrade.
    
    Thanks for your help,
    
    Joe

    Hi,
    
    Yes - the shared folder problems are indeed fixed in V3.0
    
    Cheers
    
    Stuart