<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
		>
<channel>
	<title>Comments on: SQL and Relational Theory</title>
	<atom:link href="http://antognini.ch/2009/03/sql-and-relational-theory/feed/" rel="self" type="application/rss+xml" />
	<link>http://antognini.ch/2009/03/sql-and-relational-theory/</link>
	<description></description>
	<lastBuildDate>Fri, 27 Jan 2012 07:31:12 +0000</lastBuildDate>
	<generator>http://wordpress.org/?v=2.9.2</generator>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
		<item>
		<title>By: Christian Antognini</title>
		<link>http://antognini.ch/2009/03/sql-and-relational-theory/comment-page-1/#comment-2654</link>
		<dc:creator>Christian Antognini</dc:creator>
		<pubDate>Fri, 01 May 2009 00:44:37 +0000</pubDate>
		<guid isPermaLink="false">http://antognini.ch/?p=156#comment-2654</guid>
		<description>Hi James

&gt; Your thoughts? Would it be better to disallow nulls and just use the dummy value with a histogram?

IMHO there are many situations were NULLs are useful. In fact, I think that using &quot;dummy values&quot; like 21-Dec-4000 is really not much better. Hence, the only other possibility is to design the tables to avoid such cases. But, hey, how we would model one (or several) table(s) to store &quot;first name&quot;, &quot;last name&quot; and &quot;middle name&quot; (knowing that the latter might not be available)? Of course I can think of a comple of solutions... But, honestly, I don&#039;t really like them.

In summary, I think that working without NULLs is not the way to go (at least with a database engine like Oracle). That said, the developers must be aware of the problems related to them.

Cheers,
Chris</description>
		<content:encoded><![CDATA[<p>Hi James</p>
<p>> Your thoughts? Would it be better to disallow nulls and just use the dummy value with a histogram?</p>
<p>IMHO there are many situations were NULLs are useful. In fact, I think that using &#8220;dummy values&#8221; like 21-Dec-4000 is really not much better. Hence, the only other possibility is to design the tables to avoid such cases. But, hey, how we would model one (or several) table(s) to store &#8220;first name&#8221;, &#8220;last name&#8221; and &#8220;middle name&#8221; (knowing that the latter might not be available)? Of course I can think of a comple of solutions&#8230; But, honestly, I don&#8217;t really like them.</p>
<p>In summary, I think that working without NULLs is not the way to go (at least with a database engine like Oracle). That said, the developers must be aware of the problems related to them.</p>
<p>Cheers,<br />
Chris</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: James Park</title>
		<link>http://antognini.ch/2009/03/sql-and-relational-theory/comment-page-1/#comment-2281</link>
		<dc:creator>James Park</dc:creator>
		<pubDate>Tue, 21 Apr 2009 13:07:35 +0000</pubDate>
		<guid isPermaLink="false">http://antognini.ch/?p=156#comment-2281</guid>
		<description>He does give another example on p.239 on how nulls can be problematic

My problem is that I find them useful in some circumstances. With an effective end date, for example (slowly changing dimensions), is it better to allow nulls (any current record will have an undefined end date?), or to use a dummy value like 31-Dec-4000?  

Jonathan Lewis seems to prefer allowing nulls, in p. 124 of CBO Fundamentals, but suggests histograms if nulls aren&#039;t allowed, on p. 178, so seems to allow for avoiding nulls -- but the implication is that nulls aren&#039;t a bad thing (at least from a performance perspective).

Your thoughts? Would it be better to disallow nulls and just use the dummy value with a histogram?</description>
		<content:encoded><![CDATA[<p>He does give another example on p.239 on how nulls can be problematic</p>
<p>My problem is that I find them useful in some circumstances. With an effective end date, for example (slowly changing dimensions), is it better to allow nulls (any current record will have an undefined end date?), or to use a dummy value like 31-Dec-4000?  </p>
<p>Jonathan Lewis seems to prefer allowing nulls, in p. 124 of CBO Fundamentals, but suggests histograms if nulls aren&#8217;t allowed, on p. 178, so seems to allow for avoiding nulls &#8212; but the implication is that nulls aren&#8217;t a bad thing (at least from a performance perspective).</p>
<p>Your thoughts? Would it be better to disallow nulls and just use the dummy value with a histogram?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Christian Antognini</title>
		<link>http://antognini.ch/2009/03/sql-and-relational-theory/comment-page-1/#comment-2187</link>
		<dc:creator>Christian Antognini</dc:creator>
		<pubDate>Sat, 18 Apr 2009 08:48:18 +0000</pubDate>
		<guid isPermaLink="false">http://antognini.ch/?p=156#comment-2187</guid>
		<description>Hi Peter

Thank you very much for your thoughts! It&#039;s really nice to know that I&#039;m not the only one feeling like that ;-)

Cheers,
Chris</description>
		<content:encoded><![CDATA[<p>Hi Peter</p>
<p>Thank you very much for your thoughts! It&#8217;s really nice to know that I&#8217;m not the only one feeling like that ;-)</p>
<p>Cheers,<br />
Chris</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Peter Robson</title>
		<link>http://antognini.ch/2009/03/sql-and-relational-theory/comment-page-1/#comment-2106</link>
		<dc:creator>Peter Robson</dc:creator>
		<pubDate>Thu, 16 Apr 2009 10:01:45 +0000</pubDate>
		<guid isPermaLink="false">http://antognini.ch/?p=156#comment-2106</guid>
		<description>Interesting to read these comments, as I was one of the O&#039;Reilly reviewers for this text book. Your comments on Tutorial D reflected mine exactly. Indeed, I had suggested to Chris that he cut out all that Tutorial D stuff, and write another book entitled &#039;Tutorial D and Relational Theory&#039;. He was not persuaded...

The issue of nulls is much more important. Its a pity that more space was not devoted to the subject, because, ultimately, if nulls are adopted within a relational model (and thus one departs from two-valued logic), the fundamental integrity of the database is compromised. With 2VL, one can demonstrate categorically the integrity of the results obtained from any query. Departing from 2VL, one cannot. Now transpose that uncertainty into a database system providing answers to life-critical systems, like designing aircraft. Need one say more?</description>
		<content:encoded><![CDATA[<p>Interesting to read these comments, as I was one of the O&#8217;Reilly reviewers for this text book. Your comments on Tutorial D reflected mine exactly. Indeed, I had suggested to Chris that he cut out all that Tutorial D stuff, and write another book entitled &#8216;Tutorial D and Relational Theory&#8217;. He was not persuaded&#8230;</p>
<p>The issue of nulls is much more important. Its a pity that more space was not devoted to the subject, because, ultimately, if nulls are adopted within a relational model (and thus one departs from two-valued logic), the fundamental integrity of the database is compromised. With 2VL, one can demonstrate categorically the integrity of the results obtained from any query. Departing from 2VL, one cannot. Now transpose that uncertainty into a database system providing answers to life-critical systems, like designing aircraft. Need one say more?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Asif Momen</title>
		<link>http://antognini.ch/2009/03/sql-and-relational-theory/comment-page-1/#comment-1432</link>
		<dc:creator>Asif Momen</dc:creator>
		<pubDate>Tue, 24 Mar 2009 05:29:33 +0000</pubDate>
		<guid isPermaLink="false">http://antognini.ch/?p=156#comment-1432</guid>
		<description>I do agree that the basic foundation is the key for success. I was reading Dan Tow&#039;s book and came across the classic quote &quot;Well begun is half done&quot;.</description>
		<content:encoded><![CDATA[<p>I do agree that the basic foundation is the key for success. I was reading Dan Tow&#8217;s book and came across the classic quote &#8220;Well begun is half done&#8221;.</p>
]]></content:encoded>
	</item>
</channel>
</rss>

