[r-t] Extension

[Philip Earis]

Aaargh.  Mark, a year on the Methods Committee and you seem to have become
institutionalized...

Your message is an illustrative example of how good intentions, coupled
with an excess desire to categorize (as Dr Barnes sagely points out),
coupled with a dogmatic fixation on making rules (as Robin does with
grasping onto place-notations...Roddy wisely points out the risks) leads
to an ever-lengthening mess.

Specifically, here you take a broad principle "There is a piece of work in
the extension if and only if it is in the parent" that sounds a fair basis
to work from.  Quickly you realise the limitations though, but rather than
accept the principle isn't fit for purpose you try to add a refinement
(masked in technical language), saying we can violate this principle, by
decreeing the parent contains zero examples of a piece of work.

And soon, you'll get into more and more patches and value judgements and
arbitrary stipulations and contradictions, each requiring an
ever-lengthening codifying, and all the while ever-fewer people support
this.

It's just like introducing the concept of epicycles to patch up a
geocentric model of the universe.

That's why - of all the subjects the Central Council has ever deigned to
make wordy "Decisions" on - almost a quarter of the Decisions still relate
to extending methods. It's mad.

One simply can't put together a short, succinct, self-consistent extension
system that is broadly applicable.  It's not just esoteric methods that
will show the absurdities of this - plain bob, grandsire, little bob,
little penultimus are all conceptually very simple...and yet I'd challenge
you to come up with an codified extension system that doesn't have to
carve out special cases here.

Even Robin inadvertently highlights the craziness when he admits his
current "notation-based" approach (which should be very easy to produce
algorithms to show "valid extensions of") is very difficult to do just
this basic task. So there is no simple app on the Methods Committee
website...much better for the proles to supplicate before the experts and
guardians of method purity for answers instead.

Please just accept that no one unifying extension framework is possible,
and that ringing (and the CC, and the Methods Committee) gains nothing
(and loses a lot) by trying to pretend this isn't the case.

It would be much better to delete the whole lot of current Decisions here.
 By all means publish on the website some advisory guidelines that people
can use if they really want to generate an "extension", but don't try to
mandate a broken system as something that has to be followed by all.