[Sabardeyn]

Tarisea,
Warning: this is a bit long-winded and I hope I don't seem like I'm talking down to you. It is not my intent.

I think a part of your confusion here might be terminology. A book's format is not synonymous with it's edition. The format only consists of the book's physical form and, to a minor degree, layout. The edition is the content and language only, it can be applied to multiple formats.

ISBNs are issued at the intersection of a specific edition and format. Change either portion and the ISBN must be changed.

The point here is that if you are a book collector and have multiple copies of the same book, say Agatha Christie's Murder on the Orient Express, you could enter all of them without problem into calibre as long as you differentiate each title slightly. So it is perfectly acceptable to have: 1st Edition, Signed by Author (limited print run, publisher x); 1st Edition, Signed by Author (publisher y); 1st Edition Mass Market, etc. (In these examples, the info in parenthesis is in other columns.)

Part of what you might be thinking of is a relational database approach which is part of the normalization process and basically means you try to never repeat data in any form in any field, ever. Relational database structures try to isolate unique parts of the data to be saved into separate files. But you end up with lots of little databases.

I mention the relational database info here because what you seem to want to do is put most of the metadata (title, author, tags and other columns) into a separate file. Now, in another file, you want the ISBN, the book's physical format, etc. So you could do something like:

Agatha Christie
   |
   +---Murder on the Orient Express
   |       |
   |       + Metadata: 1st Edition, Publisher X
   |       |       + Format: Hardcover
   |       |       + Format: Mass Market
   |       |       + Format: Braille
   |       |
   |       + Metadata: 1st Edition, Publisher Y
   |       |       + etc
   |      etc
   |
   +---Next Book
   |       |
   |       + Metadata
   |       + Formats
  etc

In this manner you could have all the metadata fields apply to every format you have in your records for a specific book title. But you could have lots and lots of formats. As a matter of fact, you could conceivably create a database with every edition and format ever produced for a specific book title.

Note that separating the book data into separate files as required in a relational database does not "lose" any info. It's basically a series of lookup tables with lots of cross-references. When you want to see a book's record, it just does a quick lookup, gathers all the necessary data and displays it as though it was all one record and easy-peasy to display.

Calibre's current structure does not allow relational databases (they're tough to design and use). So what you have to do is create a separate record for each relevant variant you want to keep track of. So assuming you want to keep your hardcover copy separate from your mass market, you are going to need to enter them as two separate books, probably copying most of the metadata into both book records.

I really hope that this explanation makes logical sense and is of help to you.