L DECUS U. S. Chapter     SIGs Newsletter,   Volume  5, Number 2  October 1989>                         Wombat Examiner,   Volume 11, Number 2    @ ----------------------------------------------------------------+                             Election Notice @ ----------------------------------------------------------------  H The policies and procedures of the DATATRIEVE/Fourth Generation LanguageH Special Interest Group requires that  the  SIG Chair must be (re)electedH every two years.    The  current  chair,  was  elected  at the Fall 1987H Symposium and took office at  the  Spring  1988 Symposium.  An election,H therefore, will be held at the  Fall 1989 Symposium in Anaheim; the nextE two-year term will begin at the Spring 1990 Symposium in New Orleans.   H Don Stern has  declared  his  candidacy  for  re-election  as SIG Chair.H Other nominations may be make  via  DCS  or  at  the open meeting of theH SIG's Steering Committee on Sunday, November  5, 1989, at 6:00 PM in theH DTR/4GL SIG Suite at the  Marriott  Hotel  in Anaheim.  Nominations willH also be accepted at that time for the position of Vice-Chair of the SIG.  H The elections will  be  conducted  at  the  meeting  of the SIG SteeringF Committee on Friday, November 10, 1989, at 4:00 PM, also in the Suite.    A ----------------------------------------------------------------- ,                            From the ChairmanC     Donald E. Stern, Jr., SIG Chair, Warner Lamber Co., Milford, CT A ----------------------------------------------------------------- H The symposium at  Anaheim  is  rapidly  approaching.    The SIG SteeringH Committee is  working  very  hard  to  make  the  symposium an enrichingH experience for all DECUS members  and particularly those members with an: interest in DATATRIEVE and other 4th Generation Languages.  H As usual the SIG will be hosting  a suite at the Marriott Hotel in whichH we will have Digital hardware and  software.   Come to the suite to findH an answer to a problem,  volunteer  to get involved in DECUS leadership,) or just to escape the rigors of symposia.   H Additionally, the SIG will be  sponsoring  a  variety of activities in aH campground area.  Some of  these  activities include meetings of the 4GLH Working Groups.  Members  interested in Accent-R, Corvision (ApplicationH Factory), Focus, Ingres, Oracle,    Powerhouse,  RALLY, or Smartstar areH encouraged to come  to  the  working  group  meetings  and get involved.H Users of other 4GLs are  invited  to  come  seek out users with a common	 interest.   H I'm looking forward to this  symposium  in order renew acquaintances andH to make new friends.  I'm always looking for new volunteers as well.  If8 you would like to get more involved, stop by and see me.  A ----------------------------------------------------------------- 2                      Volunteers Needed at Symposia=           Harry J. Miller, Volunteer Coordinator, Ontario, CA A ----------------------------------------------------------------- H Enhance your enjoyment  of  the  Anaheim  Symposium  by participating inH volunteer SIG Activities.  Session  chairs and suite hosts/hostesses areH need to assist with SIG  activities.  Volunteers receive an appreciationH gift of a much sought after Wombat Polo shirt.  To participate, attend aH drop-in  meeting  of  volunteers  between  5:00  and  6:00PM  on Sunday,H November 5, in the DTR/4GL SIG Suite in the Marriott Hotel (check in theH lobby for the room number)  or  see Harry Miller, Volunteer Coordinator,H at the Sunday evening Welcoming  Reception  (9:00  to 10:00PM).  You mayH also contact Harry Miller by phone at 714-988-6481 extension 7798 at theH Ontario California Police Department during the week before the Symposia8 if you would like to reserve a particular session chair.  H Session chairs have the best seat  in  the  room - right up front!  TheyH introduce the speaker, control  the  question  and answer session at theH end  of  the  talk,   evaluate   the  presentation,  enforce  the  DECUSH commercialism policy, and assist the  speaker with the lights and audio- visuals.  H Suite  hosts/hostesses  welcome  attendees,  help  direct  attendees  toH Digital engineers and experiences users to get their questions answered,. and make sure the hardware doesn't spout legs.  H In addition to the Wombat Polo  shirt,  the  SIG will also send a "thankA you" letter to the volunteer's boss if the volunteer requests it.   A ----------------------------------------------------------------- :             RALLY Working Group, Fall '89 Symposium PrimerA Steven C. Fredrickson, Chairman, RALLY Working Group, Seattle, WA A -----------------------------------------------------------------   H Well, that time is upon us  again.   The DECUS Fall Symposium, this yearH being held in Anaheim, is shaping  up  to  be another robust one for theH DTR/4GL SIG.  As chairman of the RALLY Working Group, I'd like to inviteH all of those users  and  abusers  of  RALLY,  as  well  as those who areH interested in learning about the  product,  to attend the various events scheduled for the week.   H At the time of  publication,  the  preliminary schedule was fairly well-H rounded.  The  schedule  is  outlined  below.    I'd  like to offer some highlights.    TUESDAY - November 7J  2:00 -  3:00 pm DT074 RALLY Technical Overview         (DEC Presentation)J  3:00 -  4:00 pm DT073 RALLY Tutorial Forms/Reports     (DEC Presentation)J  4:00 -  5:00 pm DT075 Optim. Perf. w/RALLY & Rdb/VMS   (DEC Presentation)    THURSDAY - November 9 K 12:00 -  1:00 pm DT061 RALLY - A Case Study             (User Presentation) J  1:00 -  2:00 pm DT078 RALLY with 3GLs                  (DEC Presentation)L  2:00 -  3:00 pm DT048 RALLY User Panel                 (Users Presentation)F  3:00 -  4:00 pm DT047 RALLY Working Group              (Redondo Room)O  4:00 -  6:00 pm DT049 RALLY/TEAMDATA Clinic            (Problem Solving Time!) #  8:00 - 10:00 pm DT005 Wombat Magic   A Tuesday,  November  7th,  the  sessions  are  being  presented by A Digital.  The first session  is  geared for those just looking at           A the product.   The  following  two  sessions provide some helpful / hints and insight into application development.   A Thursday, November 9th, there  is  a  full afternoon of sessions. A The first RALLY session, at noon,  is given by a speaker from the A Department of  Transportation.    Their  application  is a rather A interesting one, and  the  presentation  should offer some useful A insights.  Following at 1:00 pm  is a session given by Digital on A RALLY and 3GLs.   If  you're  looking into integrating RALLY with A 3GLs or vice-versa, this session is a must.  The Panel Discussion A at 2:00 pm brings  together  various  RALLY  users who will share A with the world their actual  (and often candid) viewpoints on the A product.  Come and  hear  the  perspective  from  the 'man on the 	 streets'.   A From 3:00 - 4:00 pm is  the  Working Group meeting in the Redondo A Room.  This is the opportunity to meet other users of RALLY along A with Digital personnel and  discuss  in  an open forum your ideas A and concerns.  I highly  recommend this meeting for current RALLY A users.  An  added  bonus  this  year  is  that the RALLY/Teamdata A Clinic is scheduled right  after  the Working Group Meeting, from A 4:00 - 6:00 pm in  the  DTR/4GL  SIG  Suite.   The Clinic is your A chance to  go  one-on-one  with  either  a  Digital  developer or A experienced  RALLY   user   and   get   some   answers  to  those A tough-to-solve problems.  Last,  but  not least, is Wombat Magic. A Got a neat RALLY trick?    This  Session  is just for you.  Other A users like yourself get together  to  share some 'magic' and good ; times.  There may even be a prize for the best RALLY Magic!   A All in all, Anaheim looks  like  another fun and exhaustive time! 6 We hope to see you there, so until then ... RALLY Ho!!  A ----------------------------------------------------------------- 9        Notice to all VAX/FOCUS users, and potential users 8         Les Hulse, FOCUS Working Group Chair, Boston, MAA -----------------------------------------------------------------   A A session sponsored by  all  of  the  4GL working groups entitled A "User Panel Comparison  of  4GLs"  will  be  presented on Monday,-A November 6, from 11:30AM  to  1:00PM.   This session will present A user experiences  in  implementing  different  4GLs  in different-A environments  and  will   serve   as   a  springboard  for  otherT/ 4GL-oriented sessions throughout the symposium.e  A For all users interested in FOCUS on the VAX, there will be eight A sessions and a  working  group  meeting  at  the  fall 1989 DECUSpA Symposium (November 6-10).  Six of the sessions will be sponsored,A by the DATATRIEVE/4GL SIG  and  one each by Business Applicationsh and Office Automation SIGs.g  A A working group meeting is scheduled for 9-10PM on Monday eveningcA in the Redondo Room.    A  key  discussion  topic will be how theaA working group will  approach  the  4GL  problem  presented in the A September 1989 DECUS SIG Newsletter.  This meeting is open to all           A symposium attendees with  an  interest  in  FOCUS  (who are stillo/ awake at the end of the first day of sessions).e  A A "FOCUS Wish List" will again be available in the DATATRIEVE/4GLtA campground  for  user  submissions.    IBI  will  respond  to thetA requests in the  "Wish  List"  session  on  Thursday evening from-A 8-9PM.  We hope to have  enough  time to also take questions fromh
 the floor.  @ ----------------------------------------------------------------8         Pre-Symposium Report of the Cortex Working Group>    Eric S. Dubiner, Cortex Working Group Chair, Wilmington, DEA -----------------------------------------------------------------tA The  Cortex  Working  Group  will  once  again  be  meeting  as aaA Birds-of-a-Feather session in  Anaheim.    Check the Update.Daily4  BOF schedule for time and place.  A Since Atlanta, I have  been  working  with Cortex, and their usereA group CORUS, to create a direction for the working group.  I haveiA been told that  we  will  have  at  least  one person from CortexaA Product Engineering  present  at  the  working  group  meeting toy) answer questions and respond to comments.f  A We have an opportunity to present to Cortex our concern as to theeA the  direction  of  the  product  ...  so  come  prepared  for  at productive working meeting.a  A If you have any questions  or  would  like  to be included on theeA future mailings of the Working  Group,  please do not hesitate tovA contact me at the address  given  in  the list of SIG officers inw the back of the newsletter.'  A -----------------------------------------------------------------d.                   Accent R Tips and Techniques5             Donna E. Lehman, ACCENT R Instructor, NISiA ------------------------------------------------------------------ @SCN_READ_KEYSTROKE-  A When using @SCN_READ_KEYSTROKE in a PM or CM, it will be executed A the number of times which you  indicate a value to be compared to- its result:   0          LEAVE IF @SCN_READ_KEYSTROKE EQ 278, 13  A will require two RETURNs to  leave  this  line.  It will read thetA first key and if it isn't PF3,  will wait again for the next read A to see if it's 13 (a RETURN).   If this is causing delays and thea@ necessity for repeated keystrokes, rewrite the code as follows:   (          @SCN_READ_KEYSTROKE TO @INTEGER%          LEAVE IF @INTEGER EQ 258, 13a  A Now the  keystroke  will  be  read  once  and  the  value checked 
 immediately. o   SCREEN PUT_CHARS  A If you indicate  @SCN_ERASE  in  a  SCREEN PUT_CHARS command, alleA characters on that line  will  be  erased  before the new text isyA displayed.  Specify @SCN_NOERASE if you want to keep the originali' text in addition to the new characters.b    @FILL_ACTION SETS @SCN_TERM_CODE  A If you set "END" to @FILL_ACTION,  be aware that SMF will put theeA value of  @SCN_END_CODE  into  @SCN_TERM_CODE  before exiting theeA FILL  FORM.      If   you   use   PF3   to   terminate   (258  TO A @SCN_CANCEL_CODE), and  your  PERFORM  ALWAYS  AFTER routine setsiA "END" to @FILL_ACTION,  @SCN_TERM_CODE  is  not  going  to be 258iA (PF3) after the  FILL  has  ended,  but  will  be  26 (CTL-Z), or 2 whatever the value you have set to @SCN_END_CODE.    DATA INDEX ON MULTIPLE FIELDSe  A If you usually sort  and  search  on  more than one field, definet- them as a combined domain in the data index:    %          DOMAIN ST_CITY ON STATE,CITYb  0 To select, a semicolon is an "AND" condition...   >          FIND DOMAIN ST_CITY WHEN STATE;CITY EQ 'OHIO';'PARIS'   and is faster than  4          FIND IF STATE EQ 'OHIO' AND CITY EQ 'PARIS'  
 CODE SEGMENTSG  A RELATE statements can now  be  specified  in Code Segments (after-A version 10.09) to be included in PMs.  Any code may be written inFA Code Segments and included where it would be typed into a CM, DI,   PM, SD, SF or ID, not just PMs.   
 @AUX AND @EOFa  A @AUX is usually used to determine  if a record was retrieved fromeA a data set (e.g., LEAVE IF  @AUX  NE 'YES').  It's not the entiredA or only solution to  the  last  or bad record situation, however.eA In addition, consider @EOF EQ  'YES'  to indicate end of file anda          A check the value of @AUX to  determine what really caused the @AUX A NE 'YES'.  Be sure to label @EOF - there's one for each open data0 set. 0            START:10n&           GET FROM X1 NEXT RECORD HUSH#           LEAVE:10 IF @AUX NE 'YES'a            PERFORM DISPLAY_FIELDS          REPEAT:10)          EXIT ROUTINE IF @EOF:X1 EQ 'YES'D.          PERFORM FAILED_OPEN IF @AUX EQ 'OPEN'  A or check for @AUX's other values (CLOSE, DUP, EMPTY, LOCK, MISSI,i< etc.). The "last record" read might not be the last record.    Quiz  $ What is the result of the following?   *USE DS DEC   ) *SELECT WITH NISINC ON VAX SHOW ACCENT(R)0 _m _0 _R _Y     The DBL is as follows...  # SD NISINC:                  SD DEC:L) 0001  VAX,C,2               0001  VAX,C,2i5 0002  R,I,2                 0002  ACCENT,C,6,OCCURS 4s  # DS NISINC:                  DS DEC:e2 X1 2                        X1 A     B     C     D2 X2 1                        X2 E     F     G     H2 X3 4                        X3 P     Q     R     S2 X4 1                        X4 T     U     V     W    A -----------------------------------------------------------------e)                        Dear Wombat Wizard A -----------------------------------------------------------------  Dear Wombat Wizard:s  A We deal with data which can have  a large dynamic range so we useeA USAGE IS REAL data type.    Sometimes  when  we store data into agA record we can not retrieve that record with an EQUALS RSE (recordLA selection expression).  The  problem  most  often occurs with thes          A values 0.3 and 0.6, but I  suspect that other values will exhibit A the anomaly as well.  Why can't we retrieve the data this way and % how can be work around this problem? i  .                                        Signed,  :                                        Have a REAL Problem   Dear REAL Problem:  A This is a REALly interesting  problem;  it  is one related to thedA representation of the data; and  similar problems occur in almostrA every computing language.  However, this particular one is rather  unique to VAX-DATATRIEVE.L  A So that everyone can  follow  the  details, I'll construct a testtA domain upon which we can show this particular problem.  Consider:i  1 DEFINE DOMAIN TEST USING TEST-RECORD ON TEST.DAT;a DEFINE RECORD TEST-RECORD USINGn 01 TEST-REC.    03 REAL-FIELD USAGE IS REAL.  ;   A Then use the following procedure  to  put some test data into theg domain.    DEFINE PROCEDURE LOADIT  DEFINE FILE FOR TEST;o READY TEST WRITE DECLARE X PIC 9V99.e X = 0.0  REPEAT 100 BEGIN   X = X + 0.01!   STORE TEST USING REAL-FIELD = Xh   ENDR
 END-PROCEDURE   : Then, see which records we can retrieve with the procedure   DEFINE PROCEDURE TESTITe! DECLARE COUNTER USAGE IS INTEGER. ( DECLARE X PIC 9V99 EDIT-STRING IS 9.99 . READY TEST SHARED READ X = 0  REPEAT 100 BEGIN   X = X + 0.01
   COUNTER = 0r   FOR TEST WITH REAL-FIELD = X     COUNTER = COUNTER + 1A              IF (COUNTER EQ 0) THEN PRINT X   END-
 END-PROCEDURE    When we execute TESTIT, we get   DTR> :TESTIT    X   0.15 0.23 0.27 0.30 0.43 0.46 0.49 0.53 0.54 0.59 0.60 0.67 0.73 0.86 0.92 0.98 0.99  A Thus, we find seventeen numbers which  we stored which we can not A retrieve with the same value we used  in the STORE!  If we changelA the data type on the  field  REAL-FIELD to USAGE IS DOUBLE, USAGE A IS G-FLOATING, USAGE IS H-FLOATING,  or  PIC 9V99 we can retrievetA all of the records.  However,  if  we try the procedure TESTIT innA one of the version of  DATATRIEVE-11  we can retrieve none of thee	 records!!i  A The root of the problem is how data is stored and compared in onesA or more of the many possible pairs of data types.  Floating pointtA data types such as USAGE IS  REAL are represented and stored as a.A signed fraction (a number between -1  and +1) which is called the A mantissa times a  power  of  two  which  is  called the exponent.SA [More precisely, a REAL  datum  is four contiguous bytes starting A on an arbitrary byte boundary.   Bits are labeled from the right,tA 0 through 31. The form  of  a  REAL datum is sign magnitude, withaA bit 15 the sign bit, bits 14:7 an excess 128 binary exponent, andbA bits  6:0  and  31:16  a  normalized  24-bit  fraction  with  thei: redundant most significant fraction bit not represented.]   A Thus, the representation of any  number  which  is not the sum of-A powers of twos is  an  approximate  one  when  stored in a finite A number of bits.  We have exactly the same problem in representingr2 one third in finite number of decimal digits as in          (                         1/3 = 0.33333...  > When DATATRIEVE encounters a record selection expression like   '          WITH REAL-FIELD = STRING-VALUEA  A DATATRIEVE tries to convert  the  value  on  the left side of the A equals sign and the value of the right side of the equals sign toeA a common data type with equivalent  precision.  This works if theiA right side is a STRING  data  type  and  the left side is DOUBLE,aA G-FLOATING, H-FLOATING, or  STRING,  but  sometimes  fails if theyA left side is REAL.   However,  if the Boolean comparison involvesi" ANY KIND OF A CALCULATION such as   *          WITH REAL-FIELD = CALCULATE-VALUE  A both sides are converted to  H-FLOATING  data type (the type withyA the most precision and greatest dynamic range).  [It will be leftuA for the reader to show  that  if the procedure TESTIT is modifiedt by e  +          FOR TEST WITH REAL-FIELD = 1.0 * Xi  : the Boolean fails on a different and larger set of cases!]  A If you this  this  behavior  is  strange,  consider the affect ofeA trailing zeros (implied precision) in the right side string valuet with  $ DTR> find test with real-field = 0.3 [0 records found] % DTR> find test with real-field = 0.30  [0 records found] =     .         ! keep trying with more and more trailing zerosi     .c     .i3 DTR> find test with real-field = 0.3000000000000000  [0 records found] 4 DTR> find test with real-field = 0.30000000000000000 [1 record found] i5 DTR> find test with real-field = 0.300000000000000000  [1 record found]  6 DTR> find test with real-field = 0.3000000000000000000 [0 records found] 7 DTR> find test with real-field = 0.30000000000000000000d Data conversion overflow.  [0 records found]   A Clearly, at least clearly to the Wizard, this is a problem in theoA precision of the  data  in  the  EQUALS  RSE.   This problem also A should be looked at  a  bit  more carefully by the VAX-DATATRIEVETA developers see if something more can  be done to make this behavef a bit more predictably.p  A There is one fairly obvious way  (at  least, it is obvious to theiA Wizard) to force DATATRIEVE  to  use  the  same precision on bothOA sides  of  the  record   selection   expression.    This  can  be A accomplished with the FORMAT USING construct.  An example of this  would be  4     WITH FORMAT REAL-FIELD USING 9.99 = STRING-VALUE  A This would cause  DATATRIEVE  to  do  the  comparison as strings.sA However, this would not work  for  the full dynamic range of data A that could be stored in  the  real  field.    If the value on the,A right side of the  RSE  is  prompted  for,  one could format bothD
 sides like  A WITH FORMAT REAL-FIELD USING 99.99 = FORMAT *."value" USING 99.99l  A Another way would be to  use  a global variable which has exactly ; the same data type as the field used in the RSE.  Consider n  $     DECLARE TEMP-REAL USAGE IS REAL.     TEMP-REAL = *."value"Y     READY TEST)     FIND TEST WITH REAL-FIELD = TEMP-REALF  A In this case, the FIND  will  always retrieve the desired recordseA since the data conversions AT  THE  STORE and AT THE RETRIEVE are0 exactly the same.n  A One of the best features of  the DATATRIEVE is the automatic dataiA conversion that it does behind the  scenes  - out of sight of the A user.  This is especially  true  for conversion of strings to andwA from numbers and  the  conversion  of  date  data type.  However,tA there are  situations  when  the  user,  weather  a novice wombatiA breeder or a wizard, must keep in mind what is really going on attA the level of  bits  and  bytes  to  produce  the desired results. A However, if you want  to  avoid  this  problem and can afford theeA space (4 extra bytes),  you  might  want  to  use USAGE IS DOUBLE  rather that USAGE IS REAL.    .                                        Signed,            D                                        The Wombat Wizard (WDP&SH&JG)    ?       Part 3 of Spring '89 Wombat Magic will appear next month.@  A -----------------------------------------------------------------t)                         Special RALLY PIR 8         John L. Henning, Digital Counterpart, Nashua, NH:        T. Chris Wool, PIR Coordinator, du Pont, Newark, DEA -----------------------------------------------------------------   A Digital seeks feedback from  users  of  RALLY.  What improvements A would you like to  see  in  future  versions?   In particular, we A encourage you to use the  DTR/4GL SIG Product Improvement Request2A (PIR) process.   RALLY  Engineering  has  some  specific PIRs for A which we request feedback.  If  you are a RALLY user, please read A the attached lists of suggestions,  and  give us your feedback by-A voting in the PIR process.  In addition, you can discuss the fine- points of the suggestions by:-  H     - Entering a REPLY to  one  of  the  Notes on DECUServe dealing with       RALLYe  H     - Attending  Fall  DECUS  and   discussing  them  with  the  Digital+       representatives in the DTR/4GL Suite.   H DECUServe provides an electronic  conferencing system for DECUS members;H subscription  information  is  available   by  calling  508-480-3418  or	 writting t        DECUS - DECUServe Processing     219 Boston Post Road, BP02     Marlboro, MA  01752-4605  H Discussion of RALLY primarily  occurs  within the DATATRIEVE conference,H since RALLY is one of the  products  covered by the DTR/4GL SIG.  Please1 see Notes 57-62 within the DATATRIEVE conference.r  /                        * * * INSTRUCTIONS * * *s  H A "point allocation" scheme  will  be  used  to determine your likes andH dislikes for a particular product improvement request.  You have a totalH of 50 points with which to vote.  You may allocate either positively (inH favor of a PIR) or negatively (against a PIR).  The number of points youH allocate to a particular PIR  indicates  how strongly you feel about it.H In order to  assure  a  wide  range  of  choices,  however,  you may notH allocate more that 10 points (positive  or  negative) to any one PIR and9 the absolute value of the total may not exceed 50 points.W  H Use the ballot which is in the  Questionnaire Section in the back of theH newsletter.  You may copy the  ballot if you prefer rather than removingH a page from your newsletter.   You may submit your ballot electronicallyH on DECUServe (deadline  December  4);  you  may  deliver  your ballot inH person to the DTR/4GL SIG  Suite  or  to  one of the RALLY Working GroupH offices at the Anaheim Symposium (deadline November 10); or you may mail          H your ballot to Chris Wool  at  the  address  indicated on the PIR ballot8 (deadline for receipt of mailed ballots is December 15).  A NOTE: The only commitment RALLY Engineering  has is to respond in A       writing to the top  ten  PIRs  in  the overall ranking, forhA       publication in the Wombat  Examiner  and 4GL Dispatch.  TheIA       submission of PIRs by  RALLY  Engineering in no way impliesOA       or states a  commitment  by  RALLY Engineering to implementc       the proposed PIR.r   End-user features:  N RALLY developers create applications  which  are  run  by end-users.  The PIRsB below would provide additional features for application end users:   F89-1e  / Abstract:     Provide 'Undelete record' commande  N Discussion:   RALLY applications allows users to delete a record with a singleN               keystroke.  If autocommit  is  in  effect, the user might deleteN               the wrong record without  the  ability  to recover.  This optionN               would allow users to  restore  the  deleted record.  NominationsN               for what keystroke should be  bound to "undelete record" will be=               accepted on DECUServe, at DATATRIEVE note #58].i     F89-27  0 Abstract:     Require confirmation before delete  N Discussion:   The RALLY  Guide  to  Application  Development  explains  how toN               attach a "confirm before  delete"  option  to  a form.  This PIRN               suggests that the  application  definer  be  able to select thisN               feature simply by checking a  box  on the form definition.  [ForN               discussion  at  note  58:  should  the  confirmation  message be8               customizable by the application definer?]      F89-3   . Abstract:     Remember previous query criteria  N Discussion:   Sometimes the user wants to  follow  up  a query with a slightlyN               different query.   This  option  would  cause the application toN               restore the previous query criteria when re-entering query mode,N               so the user could then edit  to make changes.  [Again, DECUServeN               can be  used  to  discuss  fine  points,  such  as  whether this2               behavior should become the default.]     F89-4   D Abstract:     Allow 'next/previous record' to cross group boundaries            N Discussion:   RALLY form/reports  consist  of  "groups".   Navigation commandsN               (such as arrow  keys)  generally  move  only  within the currentN               group, which  means  that  the  the  user  should understand theN               form/report group structure.    An  alternative navigation modelN               would hide the group  structure  from  the  end user by allowingF               next/previous record commands to cross group boundaries.     F89-5>  " Abstract:     Display current mode  N Discussion:   RALLY Form/Reports can be used in various modes (browse, insert,N               update, query ...).   This  PIR  suggests that RALLY prominentlyN               display the current  mode,  to  help  the users understand theirN               current  state.    [Note:   Screen   space  would  be  required;N               suggestions about where to place the mode display are welcome on/               DECUServe at DATATRIEVE note 58.]0     F89-6n  " Abstract:     Step-by-step queries  N Discussioni:  This  feature  would  provide  the  application  user  with veryN               specific  step-by-step  instructions  when  performing  a query,N               automatically giving hints at  each  step  of entering the mode,N               moving to a  field,  typing  in  a  valid  query expression, and"               executing the query.       Report features:   F89-7T  @ Abstract:     Adjust size of form/report to current environment   N Discussion:   Sometimes people wish to define  a single form/report for use onN               different sized devices (for example,  a 24-line VT330 screen, aN               48-line DECterm window, and  a  60-line  piece of paper).  RALLYN               lets you  do  this,  but  each  form/report  has  a single "pageN               length" which is used  in  all  situations.   This can result inN               paper reports with lots of white  space, or reports that you canE               only see part of on the screen (until you scroll them).o  N               This proposal would add an  option to adjust form/reports to the-               size of the calling RALLY task.      F89-8S  ) Abstract:     Character string formatting             N Discussion:   RALLY currently provides the ability  to format numbers and dateN               strings on output, but not character strings.  This option would=               provide output formatting of character strings.t     F89-9e  + Abstract:     Set terminal width from RALLYt  N Discussion:   RALLY tasks can be of arbitrary  width, and are formatted to fitN               the environment in  which  they  are  run.    For example, a 132N               column report can be run on  an 80-column terminal, and the userN               can scroll through it.  If  that same report (and task) were runN               on a 132-column terminal (or  DECterm window), the user need not               scroll at all.  N               At this time, RALLY makes no attempt to re-size the device it isN               using.   The  proposed  feature  would  provide  some method forN               changing the device size  from  RALLY.  [DECUServe discussion ofN               details is welcome, such  as  whether resize should be automaticN               on invocation of  a  new  task,  or  done  explicitly  by an ADLJ               function.  Please enter your opinions at DATATRIEVE note 59]     F89-10  I Abstract:     Don't start parent record on a separate page from children -  N Discussion:   This option would  provide  better  widow/orphan  control by notN               starting a parent record  unless  at  least some of its children-               will also fit on the same page.g     Menu features:   F89-11  A Abstract:     Extend Menu Builder to include All-in-1 style menuss  N Discussion:   RALLY menus can look like  All-in-1  style menus today, but onlyN               after the developer edits them.  The default style is similar toN               VAX TEAMDATA.  This option  would provide another default style,"               similar to ALL-in-1.     F89-12  A Abstract:     Display menu choices according to authorized access   N Discussion:   RALLY allows the application definer to associate passwords withN               menu choices.  This PIR suggests that both selection and displayN               of menu choices should be controlled through a protection scheme0               based upon authorized user access.               Form features:   F89-13  E Abstract:     Enhance autocommit features in parent-child situations l  N Discussion:   "Autocommit" causes changes  to  be  applied  to the database asN               soon as the user  leaves  the  current  record.  This definitionN               implies that when the user moves the cursor from a parent recordN               to a  child  record,  the  parent  is  committed.   The proposedN               enhancement would defer  committing  until  the  user moves to aN               higher group, or moves  to  another  record in the parent group.N               No commits  would  occur  when  the  user  moves  back and forth7               between a parent record and its children.      F89-14  % Abstract:     Highlight current fieldo  N Discussioni:  This feature  would  allow  highlighting  of  the current field.N               [Should a highlighted current  field remain highlighted when theN               user moves the cursor  over  to  the  List  Of Values?  Opinions;               welcome on DECUServe, at DATATRIEVE note 61.]t     F89-15  ' Abstract:     Provide more action sitesC  N Discussion:   RALLY provides  a  wide  variety  of  "action  sites", where theN               application developer can  add  additional  processing or affectN               flow of control.  This  feature  would further expand the set ofN               available  sites,  for  example,  before/after  visiting record,N               before/after performing query,  after  user  presses INSERT key.N               [To discuss what specific action  sites are needed, please reply2               to DATATRIEVE note 61 on DECUServe.]     F89-16  A Abstract:     Ability for ADL to move the cursor to a given fieldn  N Discussion:   ADL procedures are able to move the cursor from one field to theN               next (or previous field).   This feature would allow a procedureE               to move the cursor directly to a specific named field.       F89-17  D Abstract:     Ability to set keypad to numeric or "application" mode  N Discussion:   Data entry applications sometimes require  the   keypad to be inN               numeric mode.  This  PIR  would  add  an  option in ADL and/or a          N               command available on to the  end  user  that changes the mode of               the keypad.      F89-18  * Abstract:     Multiple legends for a field  N Discussion:   RALLY currently allows the definer  to specify text that pops upN               whenever the user  visits  a  given  field.    This is a popularN               feature. A proposed  enhancement  would  provide  the ability toN               specify more than a single  legend  for a given field, dependingN               on the circumstances.  [For discussion: how should the legend beN               selected?  According to the mode currently in effect?  By an ADL!               procedure?  Other?]o     F89-19  : Abstract:     Option to make fields required in query mode  N Discussion:   Some applications want  to  require  that,  if  the  user does aN               query, the user must query  on  a certain field, for example, toN               reduce the number of records retrieved.  Note: a new action siteN               "before executing query" would allow  definers to get the effectB               of this feature, although by writing ADL procedures.     F89-20  = Abstract:     Option to suppress LOV validation in query modee  N Discussion:   Some applications have stricter  restrictions on data entry thanN               on query.  This option  would  skip List of Values Validation in               query mode.'     F89-21  % Abstract:     Trap broadcast messages   N Discussion:   This feature would trap  incoming broadcast messages and present)               them in an orderly fashion.      F89-22  = Abstract:     Ability to vary the text of the working messages  N Discussioni:  Whenever RALLY  applications  perform  a  long  operation, RALLYN               automatically displays a "Working"  message.   This option wouldN               allow  application  definers  to  define  what  text  should  beN               displayed, rather than just  "working",  and thereby give a more"               informative message.             Definition time enhancements:p   F89-23   Abstract:     DCL object  N Discussion:   RALLY applications can execute DCL  commands by an External LinkN               to call LIB$SPAWN.    This  option  would  allow  the definer toN               directly specify a  DCL  command  that  is  to be performed (forN               example at an action site) thus providing the convenience of not3               needing to define the external link. i     F89-24  4 Abstract:     Ability to turn off Integrity Checking  N Discussion:   RALLY applications  are  checked  for  correctness  on  a fairlyN               frequent  basis,  for  example  as  edits  are  completed  to  aN               form/report.  Frequent integrity  checking  has the advantage ofN               giving the  application  definer  feedback  about  problems in aN               timely fashion, but  for  large  applications integrity checking(               consumes significant time.  N               This proposal would allow the definer to effectively say "I'm anN               expert - don't bother checking what  I  do until I tell you to."N               Of course, the user  of  this  feature  would have to accept theN               fact that when he finally does verify his application, there may/               be a great many errors to report.h     F89-25  " Abstract:     LSE for editing ADL   N Discussion:   This feature would make it possible to use LSE and LSE templates*               when editing ADL procedures.     F89-26  ) Abstract:     Run-time variables from DMLn  N Discussion:   A powerful  feature  of  RALLY  V2  is  the  ability  to specifyN               expressions that are  evaluated  at  run-time and which restrictN               selection of records for a  data source definition (DSD) when itN               is opened by  a  form/report.    This  proposal would extend theN               feature to also be available  for DSDs opened through ADL's Data$               Manipulation Language.     F89-27  ; Abstract:     Option for the definer to turn on/off legends             N Discussion:   RALLY "legends" provide incremental  help  messages to the user,N               and  are  a  well-received  feature  for  providing  a  friendlyN               application environment.   However,  expert  users may prefer toN               save the time required for screen painting legends.  This optionN               would provide a way for the  definer  to turn off display of all               legends.     F89-28  G Abstract:     Extend RALLY UPDATE to handle renaming of database fields   N Discussion:   The RALLY UPDATE utility will automatically make certain changesN               to  all  Data  Source  Definitions  based  upon  changes  in theN               underlying database - for example,  if the field is changed fromN               text to numeric or  the  size  is  changed.    But if a field isN               renamed  in  the   database   (e.g.   EMP_NAME  is  deleted  andN               EMPLOYEE_NAME is created),  the  definer  must  edit each DSD toN               reflect the change.  This  proposal would extend RALLY UPDATE to0               handle the case of field renaming.     F89-29  $ Abstract:     Extend use of CDD/Plus  N Discussion:   RALLY makes significant use of CDD/Plus today; but there is roomN               for extension,  for  example  by  extending  impact  analysis toN               additional RALLY objects or using additional attributes that areN               defined in the  dictionary.    [Discussion  topic for DECUServe:N               what specific extensions  are  most  important?   See DATATRIEVE               note 62.]      F89-30  ; Abstract:     Provide LOV "Starts With" for definition time   N Discussion:   A new feature of RALLY V2.1  when using Rdb databases allows theN               end-user to type in the  first  few characters of a field, pressN               Gold-Select, and see  a  list  of  choices  that  start with theN               entered  characters.    This   PIR  suggests  providing  similarB               capability within the RALLY Definition Environment.   