> > and would require more extensive revision to the code than my exact
> > anchor patch which was very small.
>
> I rather do an "extensive" change to the viewer than making an
> "extensive" change to the Plucker format. Whether it really requires
> an extensive change to the viewer remains to be seen, though...

Well, the obvious thing is that now we will need to keep track of the
last-line-length for each paragraph that continuously flows to the next one.

Here's one thing that would be problematic.  In justified text, we would
need the metrics for both the last line of text of the first "paragraph" and
the first line of text of the next "paragraph".  This would mean that to
render the last line of the first paragraph, we would need to also render
the first line of the next paragraph.

> > Moreover, if one did that, one would make the multiple-term-in-para
> > search option pretty useless and misleading, whereas now it is a very
useful
> > option.
>
> Check the paragraph attribute before the search and there should be no
> change to the behaviour...

That's simple.  Yes.


However, here's a way to make for much better backwards compatibility than
my original suggestion.  Currently, I think, the viewer completely ignores
unknown function codes.  So instead of introducing an "exact anchor"
function, we could introduce an "exact offset" function.  When this function
is encountered at the beginning of an anchor (the function is not permitted
after the beginning of the anchor), that anchor is changed from a
paragraph-based anchor to an offset-based anchor.  Then the documents will
be slightly longer with the offset-information (but the user can always turn
that feature of the parser off), but will render just fine in older
viewers--except the links will go to the top of the paragraph.

A more logical way of doing it, but one requiring SLIGHTLY more code in
control.c than my patch has, would be for the offset to indicate not the
offset from the top of the doc, but the offset from the top of the
referenced paragraph.  I like this way of doing it, because the
offset-function doesn't override the old anchor function's target, but
simply specifies it more precisely.

Alex

_______________________________________________
plucker-dev mailing list
[EMAIL PROTECTED]
http://lists.rubberchicken.org/mailman/listinfo/plucker-dev

Reply via email to