[r-t] Unusual Complib composition

[Graham John]

On 11 October 2017 at 18:43, Andrew Johnson <andrew_johnson at uk.ibm.com> wrote:

>1440 Grandsire Triples from 1440 Stedman Triples by Robert M Kilby

And Dave Sulivan replied:

>> Can someone explain how this is Grandsire Triples? It has thirds
>> every twelve changes rather than the expected fourteen.

In an extended search, Complib does try and fit compositions to a
method that has a different lead length from the method in the
original composition. This frequently comes up with useful results,
such as for Little methods and Alliance methods vs Plain and Treble
Dodging methods, resulting in different length touches. There are a
few cases, like this one, where it comes up with something unexpected.
Another case highlighted recently was when a search for Julie
McDonnell New Bob Triples worked with a peal composition of Titanic
Triples, generating exactly the same rows. As Complib handles
principles like Stedman and Titanic by splitting the changes into
quick and slow work, the notation of the new method gets mapped onto
the same structure, and the result is more like another principle (or
in the latter case, the same one). While I could suppress these cases,
my preference is to improve the automated footnotes to make it clear
when a method is effectively spliced with itself, and changes are
omitted (which could nowadays be represented using an omit call).
Addition of such a footnote for the substituted composition could make
it clear that the resulting notation is disjoint when using that
method/composition combination.

Graham