Re: Philosphy of programming

From: Programmer Dude (Chris_at_Sonnack.com)
Date: 02/17/04 

Date: Tue, 17 Feb 2004 14:44:07 -0600

"Remo D." wrote:

> A pure top-down approach seems to imply a "perfect knowledge" in
> advance (I'll partition my reality in such a way that the
> different parts will only interact in predefined ways, however
> I will sub-partition them).

It doesn't have to. It can, in some cases, be grown as you explore
the problem. You start with the top "box"--"my program"--and then
divide that into the basic steps you know you need to accomplish.
Then divide those and so on.

The idea is that the approach is a tool for exploring the problem
space. If it becomes "ugly", perhaps your thinking is wrong (OR
this type of analysis isn't well suited to the problem).

> A pure bottom-up approach seems to imply a "reductionist" view (Let
> me prepare all this little pieces, I simply will need to assemble
> them at a later stage).

It's tedious making the little parts, but it's fun putting them
together and "suddenly" having a "working" program after all that
long development time. (-:

Top-down, OTOH, can provide you with a quickly "working" program,
although it'll probably be smoke and mirrors at that point.

> Maybe our (yet to discover) method for representing "models" should
> capitalize on this observation and directly suppot a way of working
> that proceeds by attempts: an initial model is created, it is
> checked against (representations of) real cases,...

You really should read Plauger's book. This is a big part of what
he's saying: use multiple tools/approaches to solve a problem. 

-- 
|_ CJSonnack <Chris@Sonnack.com> _____________| How's my programming? |
|_ http://www.Sonnack.com/ ___________________| Call: 1-800-DEV-NULL  |
|_____________________________________________|_______________________|

Relevant Pages

  • Re: List partitioning
    ... topmost reduction of a search for an optimal partition of N items into P groups where P does /not/ divide N, into searches over all ways of distributing the N items into complementary subsets of N' and N'' items, to divide in turn into P' and P'' groups respectively. ... For each choice of distribution of N items into N'- and N''-sized subsets, the further optimal partition of each of these complementary subsets can then proceed independently. ...
    (comp.lang.prolog)
  • Re: Grouping Into 1/2 Hour Time Buckets
    ... raskew wrote: ... The third posting uses the partition() function and breaks it down into 15 ... Divide by 450 instead 900 for 1/2 hour periods. ...
    (microsoft.public.access.queries)
  • Re: nee help please reinstalling xp with new partitions
    ... not sure why it would boot into the CD set up while it was in the ... maybe it will make things easier to divide it up a bit... ... I'd say to set the C: drive for at least 40 gigs or so ... There is no way to expand a partition in XP? ...
    (microsoft.public.windowsxp.help_and_support)
  • Re: OT: Why there will not be peace in the Middle East.
    ... think such threads serve only one purpose in reality (although it may not ... have been the intent): to divide us. ...
    (rec.music.beatles)
  • Re: List partitioning
    ... topmost reduction of a search for an optimal partition of N items into P groups where P does /not/ divide N, into searches over all ways of distributing the N items into complementary subsets of N' and N'' items, to divide in turn into P' and P'' groups respectively. ... For each choice of distribution of N items into N'- and N''-sized subsets, the further optimal partition of each of these complementary subsets can then proceed independently. ...
    (comp.lang.prolog)