<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Comparision on JAVAPRO International</title><link>https://javapro-en.svenruppert.com/tags/comparision/</link><description>Recent content in Comparision on JAVAPRO International</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Thu, 07 Nov 2024 10:05:34 +0000</lastBuildDate><atom:link href="https://javapro-en.svenruppert.com/tags/comparision/index.xml" rel="self" type="application/rss+xml"/><item><title>1 Billion RECORD CHallenge - AI Style</title><link>https://javapro-en.svenruppert.com/1-billion-record-challenge-ai-style/</link><pubDate>Thu, 07 Nov 2024 10:05:34 +0000</pubDate><guid>https://javapro-en.svenruppert.com/1-billion-record-challenge-ai-style/</guid><description>&lt;p&gt;&lt;strong&gt;Previously …&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;I recently &lt;a href="https://javapro.io/2024/10/17/are-ai-coding-tools-worth-it-part-1/"&gt;wrote&lt;/a&gt; about my experiences with getting AI coding tools to find bugs and write tests.  I picked five different tools. The selection criteria were simple - free or trial with no credit card needed.  I chose three tools designed to help developers and two general-purpose LLMs.  &lt;/p&gt;
&lt;p&gt;I gave them a 1000-line Java program and asked them to do three things.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Add Javadoc to explain what the code did&lt;/li&gt;
&lt;li&gt;Analyse the code for bugs and design weaknesses&lt;/li&gt;
&lt;li&gt;Create Junit testcases that would provide reasonable code coverage.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;The results were disappointing but, in hindsight, predictable.  Overall, the tools were better at creating documentation than finding bugs and, apart from one, were pretty poor at writing test cases.  &lt;/p&gt;</description></item></channel></rss>