[Python-3000] Sane transitive adaptation

[Phillip J. Eby]

At 03:26 PM 4/8/2006 -0700, Guido van Rossum wrote:
>On 4/8/06, Phillip J. Eby <pje at telecommunity.com> wrote:
> > Those are mostly libraries, not frameworks, and for the most part you're
> > not *integrating* them.  You're not trying to make Perforce store
> > information in your database, or to put a database API over Perforce, for
> > example.
>
>OK -- then I feel vindicated that the characterization of "most Python
>programmers" as integrators doesn't apply to me. I guess I'm not a
>typical Python programmer anyway. :-)

I suspect that the perspective of people like Alex, Jim, and myself is 
distorted by our being in consulting organizations (like Strakt, my group 
at Verio, and Zope corp.) where the focus is on having a toolkit that can 
be used to produce *multiple* applications, or configurable 
applications.  Frameworks, in other words.

Verio does a lot of "private label" hosting, where each client has 
significantly different business rules.  Being able to have a core 
framework whose behavior can be significantly changed by just importing a 
different ruleset is a very useful concept.  Indeed, the one nibble I got 
about possible consulting involving RuleDispatch was from another firm 
doing private label web services of a different sort; the same kind of 
needs apply.

I suspect that Zope and Strakt have had similar influences shaping their 
perception of what a "typical Python programmer" is or does, due to the 
nature of consulting work.

I doubt, however, that we are really "typical", in the sense that most 
Python programmers do not work for consulting organizations that maintain 
an organizational "stash" of tools used to create the same kinds of 
applications over and over again, with variations.  They thus don't obtain 
any economic leverage from having tools that allow extreme customization 
while retaining modularity.

So, approaches that may seem reasonable to you or other typical Python 
programmers will sometimes be objected to by folks like Alex and Jim and I, 
in that we will tend to think of a whole series of related applications 
being written with the approach, rather than just one.  It probably also 
explains why we are usually so intently interested in "meta" features and 
often pioneer them; again we're thinking about writing entire families of 
programs.

Ironically, this no longer applies to me from a career perspective; I'm not 
in that business any more.  It frequently surprises me when I realize how 
easy it is to use what I think of as cheap hacks -- but which are quite 
adequate approaches if you intend to write only one program of the kind you 
are working on.  :)

Anyway, as to Alex's "most Python programmers are integrators" comment, I 
used to think it was reasonable.  And it probably sounded reasonable in his 
organization, and in Zope's.  After all, "most Python programmers" in such 
organizations would be a reasonable assessment.  It's just not reasonable 
in the overall Python ecosphere.