offline web reading / multicolumn web browsers

[Dave Long]

> - implicitly recursively getting later pages of multi-page articles
>   (e.g. on Wired and OSNews; clearly this has to have per-site regexes
>   and crap);

Practically it needs per-site regexes, theoretically, just consistent 
markup.

Fools that we were, we were using <LINK REL=> tags to drive UI and 
prefetching in the mid 1990's.  At the time, it seemed like a good idea 
almost universally ignored by authors, but I received an email a year 
or two ago from the guy who wrote that code, saying that he was 
pleasantly surprised to see the tag starting to be used on websites in 
the wild.  Maybe frameworks and CMS systems have been able to automate 
what site creators found too painful to do manually?

(of course, per-site regexes would really come in handy for scraping 
before fetching, for cases like 
http://www.cockeyed.com/citizen/content/content.shtml describes)


> So my proposed solution is to render the web page into a
> horizontally-narrow space as usual, but then rather than navigating
> that horizontally-narrow space by seeing a small part of it and
> dragging a scrollbar up and down to see more, you slice it into
> columns, each the height of your window, and put the columns next to
> each other, left to right.  Now there is no vertical scrollbar, but
> there may be a horizontal scrollbar for very long web pages or small
> windows.

I'd like columns, with the additional proviso that each loaded page has 
its columns laminated onto the right hand side of the window directly 
after the current page.  Then the "back button" would be just another 
function of the horizontal scrollbar.

(breaking up tables shouldn't be too difficult.  If I recall correctly, 
the table spec was based on print table layout packages, not on the 
demands of on-screen layout, and so mid-table pagination is unlikely to 
involve surprises)

-Dave