On Oct 2, 2007, at 12:27, Chris Bowditch wrote:

Andreas L Delmelle wrote:


Hi Andreas,

That is not DISagreeing with me, I think (almost on the contrary). I did not mean total-fit in the sense of the implementation of the algorithm, but total-fit as to the end-result: as such, a total-fit result may precisely require a breaking-up into many smaller best- fits. A total-fit result may just be much more difficult to accomplish with a total-fit algorithm, than by a succession of best- fit loops, interrupted and resumed at regular intervals.

Hmm. I tend to agree with Simon's perspective here. The terms "total-fit" and "best-fit" refer to the implementation of the algorithms. Surely the end result of the total-fit algorithm is the same thing as the algorithm itself? The user expectation of total- fit may well be what you describe, but I'm not sure how it could be achieved using the algorithms themselves.

Reminds me how much I hate quibbling over semantics...
In any case: if the page-sequence is portrait-landscape-portrait- etc., and FOP breaks all lines as if it is a whole sequence of portrait-pages, then I would call that "no-fit" or "first-page-only- fit"... :-)

If this can only be resolved by providing the "total-fit" algorithm *all* possible pages in advance, then /that/ is what needs to happen. I understand Simon perfectly when he states that "total-fit has no point where you can interfere"... Well then, give it all the relevant info, then send it on its way, no? Using only the IPD of the first page is simply Wrong in many use-cases.

Also, my expression of "multiple best-fits" should actually be "multiple total-fits". As in: There will always be some sort of threshold, like a certain amount of content that can be fit onto, say, the first three pages.

Once this threshold is known, running a total-fit algorithm over the elements up to that point, and continuing from the next should lead to *exactly* the same result as a global (uninterrupted) total-fit.

Just a thought



Reply via email to