pl@UUNET.UU.NET (Pedro Sanchez) (01/24/91)
I was wondering of what would be the ideal granularity of Forth programs.
Certainly we can define several ways of measuring such granularity.
A possible measure could be the mean number of words in the definition of
the several words that compound a given program.
I would like to know from you guys any comments about this subject, and
what are your views regarding the following:
1. What is (should be) the mean number of words non primitive word
definitions?
2. What is (should be) the mean number of machine instructions in a primitive
word?
3. Are your numbers based on guesses, personnal belief, or actual measured
data?
I will summarize the results to the list.
Thanks.
Pedro.
==========================================================================
Pedro Luis Prospero Sanchez internet: pl@vme131.lsi.usp.ansp.br
University of Sao Paulo uunet: uunet!vme131!pl
Dept. of Electrical Engineering hepnet: psanchez@uspif1.hepnet
phone: (055)(11)211-4574 home: (055)(11)215-6492 fax: (055)(11)815-4272
==========================================================================koopman@a.gp.cs.cmu.edu (Philip Koopman) (01/27/91)
In article <9101241516.AA11654@ucbvax.Berkeley.EDU>, vme131!lsiserv1!pl@UUNET.UU.NET (Pedro Sanchez) writes: > I was wondering of what would be the ideal granularity of Forth programs. According to Miller, the magic number for anything having to do with human thinking is 7 +/- 2. So (combining with some other theories), there should be seven "chunks" per Forth word. Roughly, each Forth word should also correspond to one chunk (except multiple words used for a single control structure can be grouped as a single chunk) to factor the code appropriately. And, importantly, control structures should seldom be nested, and never more than 2 deep. Alternately, a rule of thumb among many in the Forth community is ten words inside a colon definition. Phil Koopman koopman@greyhound.ece.cmu.edu Arpanet 2525A Wexford Run Rd. Wexford, PA 15090 *** this space for rent ***