Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.forth > #10921

Re: Lego Mindstorms

Show all headers | View raw

On Apr 5, 4:07 am, Hugh Aguilar <hughaguila...@yahoo.com> wrote:
> Symtab is available here: http://www.forth.org/novice.html

Thanks.  I like the introductory paragraph on that page which has the
following errors:

> If you have a lot of memory (modern desktop
> computers do), then a hash table is a better
> choice.

Nope.  As shown in prior conversation, a hash table is equally
effective on a system without a lot of memory.  Your problem is that
despite having the Internet at your fingers, you persist in believing
that the only implementation possible for a hash table is one that
consumes more memory than "symtab."  This has been shown wrong in past
conversation.  There are many different ways one can implement hashes
and despite being told this repeatedly, you appear to believe there is
only one way.

> Given limited memory, such as for an MS-DOS
> computer (one 64K segment for the headers),
> then symtab is the best choice.

This claim of being the "best choice" is an undemonstrated claim.
Please instrument your code to track both memory use and access time.
I will then prove your claim is false with a variety of existing
algorithms that out-perform "symtab".

> Symtab assumes that there is a lot of variance
> in how often the nodes are accessed, which is
> typical for a symbol table.

Yes, this is true-- some symbols are used more frequently than
others.  However, the essential detail you (and your "invention") miss
is that a stronger property isn't just frequency, but temporality of
access.  Locality of reference is what one would want to exploit in a
symbol table over mere frequency, and which also happens to result in
simpler algorithms that perform better, with less memory, and less
CPU.  In fact, I had shown that a simple traditional linked list with
MRU behavior would usually outperform "symtab" on both memory and CPU
because it exploits the stronger property of locality of reference.

Call me crazy, but in general, Forth programmers value simplicity and
efficiency.  "symtab" provides neither over competitive algorithms.

> If all the nodes are roughly equal in popularity,
> which is typical for an associative array, then
> ASSOCIATION.4TH would be a better choice.

This is nonsensical.  A traditional hash-based associative array tends
towards O(1) performance regardless of the frequency of access.  You
can access a symbol once or a billion times, and the cost of access
will be constant.

Back to comp.lang.forth | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

Re: Lego Mindstorms Mentifex <mentifex@myuw.net> - 2012-03-27 15:30 -0700
  Re: Lego Mindstorms dirk.bruehl@usa.net - 2012-03-27 22:52 -0700
    Re: Lego Mindstorms Hugh Aguilar <hughaguilar96@yahoo.com> - 2012-03-28 04:13 -0700
      Re: Lego Mindstorms Jason Damisch <jasondamisch@yahoo.com> - 2012-03-28 09:20 -0700
        Re: Lego Mindstorms Hugh Aguilar <hughaguilar96@yahoo.com> - 2012-03-28 17:17 -0700
          Re: Lego Mindstorms Bernd Paysan <bernd.paysan@gmx.de> - 2012-04-03 01:37 +0200
            Re: Lego Mindstorms Hugh Aguilar <hughaguilar96@yahoo.com> - 2012-04-03 04:24 -0700
              Re: Lego Mindstorms Bernd Paysan <bernd.paysan@gmx.de> - 2012-04-03 21:58 +0200
                Re: Lego Mindstorms John Passaniti <john.passaniti@gmail.com> - 2012-04-04 14:04 -0700
                Re: Lego Mindstorms Hugh Aguilar <hughaguilar96@yahoo.com> - 2012-04-05 01:07 -0700
                Re: Lego Mindstorms John Passaniti <john.passaniti@gmail.com> - 2012-04-05 07:28 -0700
                Re: Lego Mindstorms Hugh Aguilar <hughaguilar96@yahoo.com> - 2012-04-06 00:01 -0700
                Re: Lego Mindstorms John Passaniti <john.passaniti@gmail.com> - 2012-04-06 19:32 -0700
              Re: Lego Mindstorms awegel@arcor.de (Alex Wegel) - 2012-04-03 22:28 +0200
                Re: Lego Mindstorms Hugh Aguilar <hughaguilar96@yahoo.com> - 2012-04-04 01:25 -0700
                Re: Lego Mindstorms awegel@arcor.de (Alex Wegel) - 2012-04-06 17:47 +0200
              Re: Lego Mindstorms Mark Wills <markrobertwills@yahoo.co.uk> - 2012-04-05 02:12 -0700
                Re: Lego Mindstorms Hugh Aguilar <hughaguilar96@yahoo.com> - 2012-04-05 07:23 -0700
                Re: Lego Mindstorms "Rod Pemberton" <do_not_have@notemailnot.cmm> - 2012-04-06 07:19 -0400
                Re: Lego Mindstorms Mark Wills <markrobertwills@yahoo.co.uk> - 2012-04-06 07:52 -0700
                Re: Lego Mindstorms BruceMcF <agila61@netscape.net> - 2012-04-06 10:21 -0700
                Re: Lego Mindstorms "Harry Vaderchi" <admin@127.0.0.1> - 2012-04-10 14:21 +0100
                Re: Lego Mindstorms Alex McDonald <blog@rivadpm.com> - 2012-04-10 09:23 -0700
                Re: Lego Mindstorms "Elizabeth D. Rather" <erather@forth.com> - 2012-04-06 08:22 -1000
                Re: Lego Mindstorms Hugh Aguilar <hughaguilar96@yahoo.com> - 2012-04-09 19:20 -0700
              Re: Lego Mindstorms hwfwguy@gmail.com - 2012-04-06 15:30 -0700
    Re: Lego Mindstorms Bernd Paysan <bernd.paysan@gmx.de> - 2012-04-03 00:46 +0200

csiph-web