Re: [tclwebtest] HTML entities should be substituted in urls

[Grzegorz A. H. <gr...@ef...>]

On Tue, Feb 25, 2003 at 09:12:02AM +0000, Tilmann Singer wrote:
> >   http://validator.w3.org/docs/errors.html#bad-entity
> 
> I don't find any information on this topic in this link,

Well, the first list item states that using the character '&'
inside an href is invalid HTML, and should be substituted with
'&'. If you go to slashdot and take a look at the source, they
use this '&' in their url, and the browser is meant to substitute
'&' with '&' in the http request. Another explanation here:

 http://www.w3.org/TR/1998/REC-html40-19980424/appendix/notes.html#h-B.2.2

I've experimented explicitly using '&' in a manual http request,
and slashdot still recognises this and parses correctly the parameter
separator. However, this doesn't happen with our aolserver: if I
request http://something?q=bla&t=bla, the server will answer
that the parameter 't' wasn't found, probably because aolserver is
detecting 'amp;t' as the next parameter.

Without the '&' our pages are invalid HTML. With it (and without
the commited patch) tclwebtest doesn't work on our servers. Maybe
using translate_entities is overkill because only a few have to be
replaced (IIRC >,< and &).

> but shouldn't links be URL-encoded instead - e.g. " " becomes
> %20 etc.? In my opinion this should be done automatically by
> the tcl http package, but it isn't done - just checked it. It
> is done automatically by a browser (checked it with Konqueror:
> http://localhost:9000/Some Url/ becomes "GET /Some%20Url/
> HTTP/1.1").

AFAIK this is the standard escape sequence for ASCII characters
below 33 and over 126, and it's an additional substitution tclwebtest
could perform, though I still haven't been biten by this case.

-- 
 Grzegorz Adam Hankiewicz, gr...@ef.... Tel: +34-94-472 35 89.
 eFaber SL, Maria Diaz de Haro, 68, 2         http://www.efaber.net/
 48920 Portugalete, Bizkaia (SPAIN)