Re: Binary file inside fruit package

[Sebastian Crane <seabass_debian@gmx.co.uk>]

Dear Tobias,

> No, that is not how it works. It is not only nice to have.
> We want the "preferred form of modification" in the package and a binary
> blob is often not.

What would you say the preferred form of modification should be for
chess opening books?

a: the collection of millions of recorded chess games, usually in the
form of gigabytes of PGN files (which are somewhat human readable, but
usually produced and edited with GUIs)

b: that collection plus a number of filtering rules for reducing that
down to a smaller number of games (for instance, only games with
players of a certain skill level)

c: the reduced collection of thousands of recorded chess games,
produced from applying the rules to the large collection

d: the binary format, 'polyglot bin file', produced from the reduced
collection and directly used by chess engines such as fruit

As you can see, there are many different stages involved in producing
the opening book, and those with different intentions may find one to
be a 'preferred form of modification' more than others.

If the modification is just to add my favourite line to an existing
opening, I might want just option 'd', and I can edit it with my chess
program. However, if I want to create the optimal opening book for
playing against grandmasters, I might want to start with option 'a'
instead to get more control.

Finally, is the free licensing of the book file itself disputed? If
so, there will be other book files with known provenance and free
licensing that we could use instead.

Best wishes,

Sebastian