Path: dp-news.maxwell.syr.edu!spool.maxwell.syr.edu!drn.maxwell.syr.edu!news.maxwell.syr.edu!postnews.google.com!not-for-mail From: neo55592@hotmail.com (Neo) Newsgroups: comp.databases,comp.databases.object,comp.databases.theory Subject: Re: Demo: Modelling Cost of Travel Paths Between Towns Date: 20 Nov 2004 12:10:00 -0800 Organization: http://groups.google.com Lines: 78 Message-ID: <4b45d3ad.0411201210.7c3381de@posting.google.com> References: <4b45d3ad.0411071807.4cbd51f0@posting.google.com> <4b45d3ad.0411151332.4acd7159@posting.google.com> <2vugcrF2pespkU1@uni-berlin.de> <4b45d3ad.0411162100.58dedf09@posting.google.com> <301c0qF2qkiqaU1@uni-berlin.de> <4b45d3ad.0411172002.362bcdc4@posting.google.com> <4b45d3ad.0411181208.4894f98a@posting.google.com> <4b45d3ad.0411191145.2b92d71b@posting.google.com> <3074fpF2tfiu1U1@uni-berlin.de> NNTP-Posting-Host: 67.98.161.158 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Trace: posting.google.com 1100981400 1453 127.0.0.1 (20 Nov 2004 20:10:00 GMT) X-Complaints-To: groups-abuse@google.com NNTP-Posting-Date: Sat, 20 Nov 2004 20:10:00 +0000 (UTC) Xref: dp-news.maxwell.syr.edu comp.databases:42070 comp.databases.object:12961 comp.databases.theory:28493 Summarization: Goal: To demo XDb2's flexibility compared to RM. In particular, to show XDb2's query ("SELECT %.age & %.name=john;") to find any thing with age and name john, written once at design-time, continues to work forever regardless of what things have to be represented later. Neo: Assume RMDM and R2D2 are two robots sent to Mars. They have similar hardware, OS and programs except for their dbs. RMDM is based on RM and uses your above initial schema as shown below. R2D2 is based on XDb2 which doesn't require user to specify a schema. Alan's initial RM schema: CREATE TABLE test (class VARCHAR2() PK, name VARCHAR2() PK, age NUMBER); INSERT INTO test VALUES (put the values here); SELECT * FROM test WHERE age > 0; ALTER TABLE age ADD color VARCHAR2(50); UPDATE age SET color = 'brown' WHERE class = 'Tree Trunk'; // Query to find things with age and name john ??? SELECT * FROM test WHERE age > 0; Neo: Initial schema/query has some errors, please correct. Now suppose RMDM and R2D2 meet a martian who doesn't have a name and his age is unknown. Here is how R2D2 represents him using XDb2: CREATE *martian.cls = thing; // Create a martian class CREATE *; // Create a thing CREATE it.cls = martian; // Classify it as a martian // Execute query, finds nothing (for now) SELECT %.age & %.name=john; Please show how RMDM deals with the above scenario (without NULLs and redundancy) starting with corrected initial schema. Alan: Since I don't know his name, I have no way of uniquely identifying him from any other Martian whose name I do not know. You don't either. Neo: In the above scenario, that is all the info available. At the moment we are only concerned with representing that which the scenario provides. Why are you trying to represent something not given or known. It is possible to represent a thing without a name or age. You just need to start with a more generic initial schema/query. Try again. Alan: I would need to create an artificial value to assign to his name, so I'll call him Neosense1. Maybe one day, he will use his real name, and I can issue a simple UPDATE statement. Neo: Artifical values for NULLs, lead to 3VL. Do you understand why CJ Date's writes "NULLs and 3VL undermine the entire foundation of the relational model". If two martians are named "Neosense1", is it true or false that they have the same name? Alan: I don't know what you will do. Neo: Me neither, we will find out AFTER your updated schema/query. Remember, your goal is to write a query to find a thing with age and name john, regardless of what RMDM has to represent in the future. Alan: I also need to add a column to handle the new attribute you just found out about (database type) Neo: No problem, please show you new initial schema and query. You don't want to retrieve RMDM from Mars just when he meets his first martian do you? Alan: INSERT INTO test VALUES(whatever they are, but name='Neotest1', age='-1') Neo: Which RM Db handles "whatever they are"? And your not serious about age being -1 are you? Artifical values for NULLs, lead to 3VL. Do you understand why CJ Date's writes "NULLs and 3VL undermine the entire fuoundation of the relational model". If two martians have the age "-1", is it true or false that they have the same age?