Re: lynx-dev HTML entities in Lynx

[David Woolley]

> 
> In a recent note, Victor Tsaran said:
> 
> > Date: Thu, 6 Aug 1998 16:17:01 -0400 (EDT)
> > 
> > I have noticed that Lynx doesn't seem to recognise HTML entities such as
> >   and alike. In fact, when sending HTTP headers Lynx doesn't strip
> > out these entities from the URL and thus causes web servers to complain
> 
> Yow!  Do you mean that some server is serving a page with entities
> in a URL?  I suppose that's legal, although I'm much more accustomed

That is perfectly legal and necessary in some cases, e.g. to generate 
a form type URL from a normal anchor (&).  If Lynx really is 
passing through entities in the HTML source code to URLs sent back to
servers, it would be broken.

My feeling is that what is being talked about here is command line
entered URLs, where one would not expect entity decoding to happen;
unlike URL encoding, the entities are not part of the URL, but part of
the HTML used to embed the URL, and there is no HTML layer involved
in command line input.

Incidentally, I would consider any site that wants non-breaking spaces
in URLs to be sailing too close to the wind for comfort, even though
I think they are considered URL safe.

> to seeing special characters in URLs encoded with the %xx construct.
> But I'm curious:  Can you provide the URL of an example of such a page?