[gclist] Re: Jones & Lins's book

[Richard Jones]

Kragen Sitaker wrote:

>I just got the Jones & Lins book ("Garbage Collection") a week or two
>ago.  I was disappointed to discover I'd gotten the November 1997
>printing, since reportedly the new printing has some errors corrected,
>including several that I found.  

If I'd only bought the book a week or two ago, I'd be somewhat miffed with my 
bookshop for selling me an old printing. The January 1999 version *does* have 
all the reported errors corrected. The errata for the older printings can be 
found at:

	http://www.cs.ukc.ac.uk/people/staff/rej/gcbook/errata.html

(the older URL printed in the book also works)

>[I note that some of these errors
>would probably have been obvious if the algorithms in the book had been
>written in a real programming language and tested before the book was
>published.]

I am not sure that this is true: very few of the errors were ones that a 
compiler would spot (and indeed, there were also errors in some fragments of 
real code). Spotting the errors would therefore have meant implementing all 
the algorithms *and* the applications to test them. This would have been an 
enormous task.

I thought long and hard about how the algorithms should be represented. In the 
end I concluded that pseudocode offered the best representation:
* it is compact
* it is more expressive
* it is not cluttered with the 'book keeping' detail required by real languages
* it doesn't lead to a programming language specific bias
* it doesn't mislead anyone into believing that the algorithms are more than 
  illustrative: real implementations of efficient collectors are necessarily 
  much larger and more complex.

>Finding the book was not easy; of course, my local bookstore didn't
>carry it, so I had to special-order it.  In the process, I discovered
>that Books in Print lists it under the wrong title (at least, a title
>that's not the same as the one printed in 100-point letters on the
>front cover) and has it categorized as being about "Refuse and Refuse
>Disposal".

Try Amazon - they usually seem to effer pretty quick despatch. The .com and 
.co.uk URLs for the book are:

	http://www.amazon.com/exec/obidos/ASIN/0471941484/richardjones
	http://www.amazon.co.uk/exec/obidos/ASIN/0471941484/richardjones0d

As for the categorisation, one can but try. My publishers, Wiley, asked for 
the sub-title "Algorithms for Automatic Dynamic Memory Management" to avoid 
this sort of thing. Certainly any search of the net for "garbage collection" 
comes up with all sorts of suggestions - hence the (rather pathetic) joke in 
the Preface.

>I'm really enjoying the book, by the way.

Glad to hear it!


Richard Jones

Computing Laboratory                                
University of Kent at Canterbury
http://www.cs.ukc.ac.uk/people/staff/rej/