[Date Prev][Date Next][Thread Prev][][Date Index][Thread Index]

idea: short circuit jumping to the next child page

I have an idea for a new command that will put emacs-w3m way in front
of the "sales competition".

Let's look at a typical browsing keypress log.
Here the user is browsing one by one all the links on a page, and each
to only a depth of one.

...SPC SPC B <tab> <return> SPC SPC SPC B <tab> <return> SPC SPC SPC SPC
B <tab> <return> SPC SPC SPC B <tab> <return> SPC SPC SPC SPC...

SPC runs the command w3m-scroll-up-or-next-url
B runs the command w3m-view-previous-page
<tab> runs the command w3m-next-anchor
RET (translated from <return>) runs the command w3m-view-this-url

We see he has to go back (B) to the parent page, move to the next link
(<tab>), click it (<return>), each time he wants to get to the next
link on the parent page.

Yes, if the "next-url" information were present in each child page,
then w3m-scroll-up-or-next-url could use it to do the jump for him.

However that relative link information is not present, as the pages
were all made by independent creators.

This is where emacs-w3m could come in to bridge the gap: save
knowledge of what the "virtual" next and previous URLs might be, and
"bust through" to them anyway at the bottom or top of a page when using
SPC runs the command w3m-scroll-up-or-next-url, or
<backspace> runs the command w3m-scroll-down-or-previous-url

So the above sequence becomes just ...SPC SPC SPC SPC SPC SPC SPC...
or perhaps ...SPC X SPC SPC SPC X SPC SPC X SPC... where X is the new
jumper command, if we fear overloading SPC, or want to provide a jump
here and now, instead of requiring the user to first go to the bottom
of a page for the magic jump of w3m-scroll-up-or-next-url to kick in.
Same is needed for the backwards direction.
Indeed, maybe a variable should be available to turn this automatic
jumping into y-or-n-p mode jumping, if the user fears one too many
SPCs might land him on a page he had no interest in going to.

Though I use this to only depth of one (here browsing WWWOFFLE
last time on line indexes), it seems there is no restriction on depth:
just remember the links above and below the current one we are
browsing, and jump to them when we wish to bust through the top or
bottom of the page we are on.

OK, I hereby hand my idea to the engineering department for further