crunch mode

#71 11-25-2004, 06:56 AM

In article <ez%od.30089$5b1.2995@newssvr17.news.prodigy.com >,
phlip_cpp@yahoo.com says...
> Gerry Quinn wrote:
>
> > [And by doing something, I mean 'achieving a useful effect', not
> > 'implementing a named pattern'.]
>
> Ah, yes, the possibility of Pattern Abuse is more than enough reason to
> refuse to achieve a useful effect using a named pattern.

A complete mis-reading of what I said.

Consider the following programming task: "lighting a cigar". You have
two languages "Matches" and "Lighter".

"Matches" implements the "LightByStriking" pattern. This pattern is
very difficult to implement in "Lighter". [You could tie a rubber band
around the gas lever and strike the friction wheel against a wall, I
suppose.]

Nevertheless, "Lighter" is just as good for the programming task of
lighting a cigar. "LightByStriking" is not in itself a useful effect, it
is just one way of achieving something that works towards a useful
effect.

- Gerry Quinn

#72 11-25-2004, 07:05 AM

In article <Pine.WNT.4.53.0411241344000.668@SLINKY>, flippa@flippac.org
says...
> On Wed, 24 Nov 2004, Phlip wrote:
> >
> > > [And by doing something, I mean 'achieving a useful effect', not
> > > 'implementing a named pattern'.]
> >
> > Ah, yes, the possibility of Pattern Abuse is more than enough reason to
> > refuse to achieve a useful effect using a named pattern.
> >
> I have to admit, I tend to be a little skeptical about patterns due to
> examples like the Visitor pattern where half the problem is that typical
> OO techniques get the dependencies backwards (the whole "pass-and-return
> context while walking a graph" thing, that's fine). Mostly it just boils
> down to my tending not to use pattern names to refer to things though -
> there're useful patterns out there. Ones like MVC I'm even willing to use
> the name for.

Often they seem to be just jargonisation of the trivial. I wonder when
they will be discovered by 'critical theorists'?

- Gerry Quinn

#73 11-25-2004, 07:05 AM

In article <Pine.WNT.4.53.0411241344000.668@SLINKY>, flippa@flippac.org
says...
> On Wed, 24 Nov 2004, Phlip wrote:
> >
> > > [And by doing something, I mean 'achieving a useful effect', not
> > > 'implementing a named pattern'.]
> >
> > Ah, yes, the possibility of Pattern Abuse is more than enough reason to
> > refuse to achieve a useful effect using a named pattern.
> >
> I have to admit, I tend to be a little skeptical about patterns due to
> examples like the Visitor pattern where half the problem is that typical
> OO techniques get the dependencies backwards (the whole "pass-and-return
> context while walking a graph" thing, that's fine). Mostly it just boils
> down to my tending not to use pattern names to refer to things though -
> there're useful patterns out there. Ones like MVC I'm even willing to use
> the name for.

Often they seem to be just jargonisation of the trivial. I wonder when
they will be discovered by 'critical theorists'?

- Gerry Quinn

#74 11-25-2004, 09:50 AM

On Thu, 25 Nov 2004, Gerry Quinn wrote:

> In article <Pine.WNT.4.53.0411241344000.668@SLINKY>, flippa@flippac.org
> says...
> > I have to admit, I tend to be a little skeptical about patterns due to
> > examples like the Visitor pattern where half the problem is that typical
> > OO techniques get the dependencies backwards (the whole "pass-and-return
> > context while walking a graph" thing, that's fine). Mostly it just boils
> > down to my tending not to use pattern names to refer to things though -
> > there're useful patterns out there. Ones like MVC I'm even willing to use
> > the name for.
>
> Often they seem to be just jargonisation of the trivial. I wonder when
> they will be discovered by 'critical theorists'?
>

In fairness they do tend to include the most general variant on the theme
- ironically, passing a context makes rather a lot more sense to FP coders
than imperative ones, a classic example being shunting an environment
around an abstract syntax tree. The Observer pattern does leave me
thinking "oh look, somebody reinvented interrupts!" though.

I find patterns far more useful implemented as HOFs and the like. A
"support for the Observer pattern" class or monad could be quite a
timesaver within even one project, and I'm probably going to end up
hacking one up for the IRC client I'm working on.

--
flippa@flippac.org

#75 11-25-2004, 09:50 AM

On Thu, 25 Nov 2004, Gerry Quinn wrote:

> In article <Pine.WNT.4.53.0411241344000.668@SLINKY>, flippa@flippac.org
> says...
> > I have to admit, I tend to be a little skeptical about patterns due to
> > examples like the Visitor pattern where half the problem is that typical
> > OO techniques get the dependencies backwards (the whole "pass-and-return
> > context while walking a graph" thing, that's fine). Mostly it just boils
> > down to my tending not to use pattern names to refer to things though -
> > there're useful patterns out there. Ones like MVC I'm even willing to use
> > the name for.
>
> Often they seem to be just jargonisation of the trivial. I wonder when
> they will be discovered by 'critical theorists'?
>

In fairness they do tend to include the most general variant on the theme
- ironically, passing a context makes rather a lot more sense to FP coders
than imperative ones, a classic example being shunting an environment
around an abstract syntax tree. The Observer pattern does leave me
thinking "oh look, somebody reinvented interrupts!" though.

I find patterns far more useful implemented as HOFs and the like. A
"support for the Observer pattern" class or monad could be quite a
timesaver within even one project, and I'm probably going to end up
hacking one up for the IRC client I'm working on.

--
flippa@flippac.org

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode