[Koha] sorting long Cutter's

[Giuseppe Angilella]

Hi,

call numbers having Cutter parts made of several substrings currently are 
not sorted correctly. Here are examples:

| itemcallnumber     | cn_source | cn_sort                        |
| 530 F435 1996 v1   | ddc       | 530_F435_199600000000000_V1    |
| 530 F435 1996 v2p1 | ddc       | 530_F435_199600000000000_V2P1  |
| 530 F435 1996 v2p2 | ddc       | 530_F435_199600000000000_V2P2  |

Despite being in the Dewey 530 class, these items are sorted after any 
530.xy item (and before any 531, which is OK).

I suspect this has to do with the additional space separators: more than 
one seem to be too many.

Any hint? What is the actual algorithm used by Koha to evaluate the 
cn_sort field (within the ddc scheme)? Grouping, or even merging, all the 
subfields after the Dewey into one single string was of no help.

Many thanks and best regards,

Giuseppe.