Re: The Second Book Of Machine Language on the Web

["Bryan Parkoff" <spam>]

Michael,

    Which machine language book is the best for advanced assembly
programming?

Bryan Parkoff

"Michael J. Mahon" <mjmahon@aol.com> wrote in message
20030701144034.09237.00002113@mb-m29.aol.com">news:20030701144034.09237.00002113@mb-m29.aol.com...
> Kevin Savetz wrote:
>
> >The team at AtariArchives.org is pleased to announce that the full text
of
> >the
> >best-selling book _The Second Book Of Machine Language_ by Richard
Mansfield
> >is now online at
> >
> >    http://www.atariarchives.org/2bml/
> >
> >Published in 1984 by Compute! Books, this classic book is the sequel to
> >_Machine Language For Beginners_, another of Mansfield�s best-selling
titles
> >which is also available on the Web site. The Second Book Of Machine
Language
> >walks readers through the creation of LADS (Label Assembler Development
> >System,) a sophisticated assembler written in machine language. It
includes
> >examples and program code for Atari, Apple 2, PET/CBM, VIC-20, and
Commodore
> >64 computers.
> >
> >This is the 17th classic computing book to be made available at
> >AtariArchives.org. Like all books at the site, it is available with the
> >gracious permission of the author.
>
> I think that this project to make useful books available on the web
> is a valuable one, and I commend those responsible.
>
> However, with regard to this particular book, I have my doubts.
> As nostalgia, it is a great contribution for all those who once
> read it and find it familar--but it can hardly be considered
> the work of a good assembly language coder.
>
> I have often said that the best way to learn to write code in a
> language is to read great code written by great coders.  This
> book does not present great code.  It might be defended as
> an introduction to coding, but it is then a _second_ introduction
> by the same author.
>
> The LADS assembler which is the focus of the exposition is
> hardly an "advanced" assembler (which would have macros
> and, at least, the ability to handle address expressions
> more complex than "label+constant").
>
> Not only are the assembler's specifications elementary, its
> construction and algorithms are elementary.  For example,
> using sequential search for the mnemonic table is a poor
> choice, given the relative ease of, say, binary searching
> in an ordered table.
>
> The choice of a sequential table of labels is, perhaps,
> intuitive to an inexperienced coder, but experience soon
> leads to the realization that a partially-ordered tree structured
> table is _much_ more efficient.  Add to this the fact that
> the author chooses to insert all labels into the table without
> searching first, so duplicates can only be detected by doing
> a _full_ search of the sequential table on _each use_, and
> you see a totally naive design realized totally naively.
>
> And the way of thinking about performance issues presented
> is also backwards, choosing to focus on coding details before
> reconsidering algorithms.  And then choosing code optimizations
> without any consideration of measurement to locate the "hot"
> regions of code.  (For example, sampling the PC on timer
> interrupts will _instantly_ reveal the location of any region
> of code which is consuming a disproportionate amount of
> execution time.)
>
> All things considered, this book could be useful to a relatively
> inexperienced machine language coder who wanted a better
> familiarity with what an assmbler does and how it does it.
> But this familiarity would be obtained along with some
> fundamental misunderstandings about how to design a
> a good program and implement it both wisely and cleverly.
>
> A minor but annoying factor throughout the text is the use
> of decimal representations of bytes, where hexadecimal
> would be more familiar (to anyone looking at memory)
> and clearer, since the patterns of binary values are
> manifest in hex representation.
>
> Many years ago, an instructor mentioned in passing that "the
> problem with most assemblers is that they were written by
> someone who is writing their first assembler."  This observation
> is confirmed by this book (as well as by much experience).
>
> (BTW, the irony of reviewing a book almost two decades
> after its publication is not lost on me.  Such a thing would
> never had occurred to me if not for its "republication".  ;-)
> -michael
>
> Check out amazing quality 8-bit Apple sound on my
> Home page:  http://members.aol.com/MJMahon/