Re: lynx-dev jumping around

[Philip Webb]

990413 Klaus Weide enlightened me: 
>> i have an URL in my bookmark file & now jumps file
>>  
>> http://www.tor.ec.gc.ca/forecasts/forecast.cgi?city=Toronto&province=Ontario
>> (of course, it's in the proper anchor setting).
>> when i use the bookmark or jumps listing, there's no problem;
>> when i enter  j  then  w  (the short-cut), i get a screen telling me
>> i haven't entered the full URL: it seems to be  www.tor.ec.gc.ca/  only;
>                                  =======================================
> What do you mean by that?  www.tor.ec.gc.ca/ is not a URL.
> I doubt that the INFO page says  URL: www.tor.ec.gc.ca/

it says (centred in the original):

  weather.ec.gc.ca

  Thanks for visiting weather.ec.gc.ca!
  [1]http://weather.ec.gc.ca/

this is NOT the page i reach via the bookmark's full URL above:
this page suggests adding the  ?  and  &  bits to get the local forecast.

>> even stranger,  =  tells me i have reached the proper destination,
>> which is a simple Toronto weather outline & forecast.
> What's the URL according to the INFO page?  Does it have '&'?

it says (with formatting):

 File that you are currently viewing

    Linkname: weather.ec.gc.ca
         URL:
           http://www.tor.ec.gc.ca/forecasts/forecast.cgi?city=Toronto&amp
           ;province=Ontario

this is a bug in Lynx: that's not the document i am viewing.
           
> The problem in this specific case seems to be the '&'.
> Yes, it's more correct than just writing '&',
> but mechanism B doesn't understand it
> and simple '&' should also work for method A.  So just change it. 

yes, that does get me to the usual weather-forecast page.
                                                 
> Shortcut files are being read by 2 completely different mechanisms,
> plus one for reading JUMPFILE from  lynx.cfg .
>   A. normal HTML parsing: treated like any other text/html file,
>      with everything that entails.  That's what is used
>      when you actually 'go to' or 'jump to' (with '?') the shortcut file.
>   B. the 'shortcut way': file is read *once* (on first use of the jump key),
>      a list of (shortcut name, target URL) pairs is built,
>      the list doesn't change until program exit.
> Mechanism B doesn't know much about HTML.
> That's why it may break if the format is too different from the samples
> (similar problem as with bookmark files).  It also doesn't know anything
> about entity unescaping, the URL strings are just used as they are found.

it's good to be enlightened, as always, but how would anyone ever know that?
even more important, why are things arranged that way?
and to repeat the PITA from my other message,
why doesn't Lynx re-read the jumps file after  ^r  when viewing the list?
it should be very simple: where should i look to make the correction?
to emphasise the point, if i now want to use  j  to get the weather,
i shall have to restart the Lynx i have running under Screen,
which already has a lot of cookie choices, passwords etc nicely stored,
which i will have to re-enter here & there.

then there's my 3rd gripe: jumps refuses to recognise short-cut names
if they aren't in alphabetical order: is there any reason for this?

this is all another example of the way Lynx was obviously thrown together
in the early days with hasty documentation aimed at sysadmins,
but has never been revised & cleaned up for more upto-date needs.
no criticism of anyone, but a need to focus attention on such things.

-- 
========================,,============================================
SUPPORT     ___________//___,  Philip Webb : address@hidden
ELECTRIC   /] [] [] [] [] []|  Centre for Urban & Community Studies
TRANSIT    `-O----------O---'  University of Toronto