[r-t] 23-spliced

[Graham John]

Ander wrote:

> My first suggestion would be to ignore the library of rung
> methods and just pick ones that work (rung or unrung).

The risk then is producing a composition containing methods which work, but
nobody wants to ring. I tried that 10 years ago with 14 atw Surprise Royal -
and guess what - it hasn't been rung.

> Other criteria might be: no more than 4 places made simultaneously

Agreed. You could also argue that the best methods have no more than 2
simultaneous places.

> no three blows in the same place

Agreed.

> no bell spending more than x blows consecutively in two adjacent places

Harder to check. You obviously don't like Cray or Derwent then!

> no backstroke 87s in the plain course

Hmmm. Not sure how I check this easily from the notation.

This is what I have filtered so far on a microsiril S8 file of a few months
ago:-

             Methods input =  4576
                   Group a =  352
                   Group b =  855
                   Group c =  258
                   Group d =  308
                   Group e =  175
                   Group f =  1292
                   Group g =  108
                   Group h =  98
                   Group j =  151
                   Group k =  87
                   Group l =  154
                   Group m =  690
               Not regular =  48
 Methods of non std length =  1
         Methods with pn78 =  526
   Methods with > 4 places =  149
   Methods with > 2 places =  2351
Methods with >3 adj places =  477
Methods with >2 adj places =  1516
    Methods with > 2 blows =  19
            Methods output =  1374

I wanted 2nds place regular methods only, so by adding a filter for more
than 2 places made, no pn78 & no 3 blows reduces it to 1374 methods. I still
need to reduce this by a factor of two to three though.

Graham