Technical Status Report - 95Q1

                                     Ian Davis
                               University of Waterloo

         Progress Report

         In the period April 1st - June 30th 1994 our primary focus was on
         development  of prototype software, capable of serving as a proof
         of  concept  for  the  paper  entitled  'Text/Relational Database
         Management Systems: Harmonizing SQL and SGML.'

         Work  on  parsing  the  relevant  sections  of the SQL-92 grammar
         (containing  our  text  extensions)  was  completed,  and work on
         supporting  the  hybrid  query  processor  catalogue  begun.  The
         abstract syntax tree produced by the 'YACC' parser was translated
         into a more useful operator tree.

         Preliminary  work  started  on performing the translation between
         this  operator  tree  (described  using  SQL  constructs) and the
         corresponding   commands   (described   using   PAT  constructs),
         necessary to interface with the existing Open Text text database.

         Preliminary  work  also  began  on  the  design  of a generic SQL
         database  engine,  and  the  functions  that it must support.  To
         facilitate  rapid  development  of  this  software, a testbed was
         developed  which  translates a low level functional language into
         an  acyclic  directed  execution  graph.   This  resulting  graph
         provides  the instructions on which the database engine operates.
         This  database  engine will later be incorporated into the Hybrid
         Query  Processor,  and will potentially also be incorporated into
         agent   processes  which  provide  SQL  extensions  not  directly
         supported by the underlying text engines.

         Because  it is our intent to provide support for structured text,
         tools  were  developed  which  allow  an  arbitrary SGML grammar,
         defined  using  the  SGML Data Definition Language, to be parsed.
         Using  these  tools  the  direct  membership of text nodes within
         other  nodes  was  computed, as was the transitive closure of all
         such memberships.

         An   ODBC   testbed  was  developed,  and  existing  source  code
         (providing  ODBC  connectivity  with  ORACLE  via  underlying OCI
         calls)  was  integrated into this testbed.  This testbed supports
         the use of Sun Remote Procedure Calls (RPC).

         Triton,  our PC clone, was reconfigured so that it operated under
         Windows 3.11.  The Fulcrum software (provided by Paul Cotton) was
         then  installed  on  this machine.  LAN-workplace for Windows was
         also  installed  on  this  machine,  providing  interconnectivity
         between this machine, and our other Unix based platforms.

         To  facilitate  the  development  of  low  level  ODBC interfaces
         between  the  Hybrid Query Processor resident on a Unix platform,
         and  Fulcrum  software resident on the PC, public domain versions
         of  the Winsock socket libraries and Remote Procedure Call source
         code  were  obtained,  and  installed  on  the  PC.  Watcom's 'C'
         compiler was also installed under Windows.


         Achievements

         The  major achievements this quarter have been the implementation
         of  an  ODBC driver, which interfaces with our Unix based version
         of Oracle; the installation of Fulcrum's text engine on a PC; and
         on   going  development  of  Hybrid  Query  processor  and  agent
         software.

         Paul  Cotton's  visit  of  May  9-10  was  constructive,  and his
         willingness  to  assist  us  much  appreciated.  He has been very
         helpful  in providing us with background material of relevance to
         this project, and in making Fulcrum software available to us.

         The  paper entitled 'Text/Relational Database Management Systems:
         Harmonizing SQL and SGML' was distributed to all members of CSSC,
         and presented last week at ADB-94, in Sweden.


         Problem review

         There  is  a need to establish a better working relationship with
         other  members  of  the consortium, and to become better informed
         about the activities of all the members within the consortium.

         We need to become more familiar with the Fulcrum text engine, and
         must   rapidly  establish  the  interfaces  needed  in  order  to
         communicate  with  this  engine.  We are assuming that it will be
         relatively  easy to establish remote procedure calls between Unix
         and  Microsoft  Windows, but this remains uncertain.  We are also
         assuming  that  the Fulcrum text engine provides some support for
         the decomposition of SGML structured text.

         Research  is  needed in order to define what constitutes the core
         or  desired  functionality  provided by an arbitrary text engine.
         Ideally,  such  research would lead to good models for describing
         important textual operations.  Further research into the needs of
         those  who  manipulate text is also warranted, in order to better
         understand the requirements of end users.


         Coming events plan

         Weekly  meetings of the research group within the University will
         continue  throughout  the  next  period.  Further meetings of the
         Hybrid  Data  Model working group (of which we are a participant)
         are  planned,  but  dates  for  these  meetings have not yet been
         finalized.