Path: utzoo!attcan!uunet!willett!ForthNet
From: ForthNet@willett.UUCP (ForthNet articles from GEnie)
Newsgroups: comp.lang.forth
Subject: All Sorts of Sorts
Message-ID: <451.UUL1.3#5129@willett.UUCP>
Date: 13 Feb 90 02:31:37 GMT
Organization: Latest link in the ForthNet chain.  (Pgh, PA)
Lines: 37


 Date: 02-11-90 (08:41)              Number: 2884 (Echo)
   To: IAN GREEN                     Refer#: NONE
 From: STEVE PALINCSAR                 Read: NO
 Subj: SORTING ALGORITHMS]           Status: PUBLIC MESSAGE

 As a matter of fact, even as august an authority as Knuth says that for 
 some sorts of data, the selection sort happens to be the best one 
 available.  But besides that, your assumption that "the real world" 
 consists only of massive databases simply takes my breath away. 

 As it happens, the application that I translated KNuth into Forth for is
 a real-world application for a task I do at work.  Every month I do a 
 search on a Library of Congress bibliographic database on their inhouse 
 SCORPIO system and I download a list of their reports added since the 
 previous issue.  (That's usually between 60 and 100 or so items.)  The 
 entries are downloaded in report number order.  My program strips off 
 the unnecessary material from each citation, and puts the cites in 
 author title order.

 In fact, I think the requirement to order the data came about because 
 one of the editors of the publication secretly wished to eliminate the 
 column, and thought that the amount of manual effort required to do so 
 would cause me to simply abandon the effort.  Ha!  Using this program it
 takes me about 30 seconds to process the data.  As it so happens, she's 
 taking over production of the column, and was astounded at how easy it 
 was...

 Incidentally, if you'll stop to consider the situation, you'll see that 
 the bubble sort is less efficient than the selection sort - so by your 
 own criteria you oughtn't be using it!  But as we both know, there are 
 times in the real world that you need to sort relatively small numbers 
 of items, and in such cases more complex algorithms may be even slower
 than the simple ones.  Even Knuth says so...  ;-)
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'