AltME groups: search

Yes.

If you clicked the link I gave above, then you saw a page served 
as text/html  [probably should be textplain -- so I've changed it]
If you try format=rss then you get a page served as text/xml


In both cases, the output is not meant for humans: one format is 
for REBOL and one for RSS readers.

I know it's now for human readed, but for example this chat is public 
and if someone would click on the link, now it looks much more better. 
Don't forget, that Rebol should be human friendly:)

yes, no problem, and the issue with the bots - if the bot don't support 
cookies (non of them does), i can give him whatever I want, I'm not 
cheeting, I just may think, that it's somethink like LYNX and serve 
him pure text pages:) And if he don't like it so it's his problem 
(or its?) And with the robot.txt file - ugly bots will not respect 
robot.txt file anyway :)

Some bots (the more evil ones) have an even more evil human at ther 
side....Those bots can handle cookies, and will also use the human 
to step them through any logon procedures.
So, technically, yes: bots can use cookies.


But on the cloaking issue: if you show the *same* content to any 
visitor that does not use cookies then that is *not*cloaking, even 
if you serve different content to those that do. So no problem there.

Kudos to Yahoo!, who today released two pieces of goodness into the 
commons. The first is their UI library, and the second is their Design 
Patterns Library. The UI Library is a collection of DHTML/Ajax/Javascript 
(pick your favourite term) controls and widgets. The Design Patterns 
Library is "intended to provide Web designers prescriptive guidance 
to help solve common design problems on the Web". 

- http://developer.yahoo.net/yui/
- http://developer.yahoo.net/ypatterns/

just run it and than I can do read pages and cookies are processed 
automatically

You mean HTML tables?

The cell has a height and width, and the image has a height and width.

You probably need to set both height and width on both the cell and 
the image.
Probably easiest with CSS 
Remember to set the padding and margin to zero.

And remember that IE6 and lower handles this differently to other 
browsers, so it's not easy to get pixel-perfect borders and so on.

I have those photos, numbered jpges, I will provide here view script 
to put comments in there, and then I want to automatically generate 
the content ....

browsers actually do a good job of pixel-perfect, but printers don't 
do pixels.  using real-world css dimensions, like cm or pt etc. will 
translate between device contexts.  Anyway.  I don't envy the task. 
 


Anyway, I have been messing with embedding REBOL in client-side code, 
which is working pretty well: http://eisic.ws/ext/r/Document2.plugin.r.html


I need to figure out how to keep REBOL from bailing out on me, though. 
 generally, if the console pops up, I have to refresh the page.  
For instance,  any print will pop up the console, and I would really 
rather not pop up the console from the page, because closing it destroys 
the REBOL instance.

Common mistake is case of the filename and the way it's write in 
the html.

It happens ofen when you test on Windows and the web server is under 
*nix.

Yes, upper and lowercase.

<pre> literally means "as written" -- if there are no line breaks 
in the text, then there will be none on the page.
***

Some lines in <pre> tags can be accidently enormously long, and need 
to be wrapped by hand.

That's one reason REBOL.org offers you a user-setting for the point 
at which you want Mailing list messages to be forcibly wrapped:
http://www.rebol.org/cgi-bin/cgiwrap/rebol/cpt-update-profile.r

(See Appearance and settings / Point at which Mailing List Archive 
messages will start to wrap.)

Dideir, yeah we ran into this recently in allowing people to escape 
code inside our Wiki.  We allow Rebol, HTML, etc.

At first it seems counter intuitive. But this is the way of HTML.


Aside from the straight forward concept as Sunanda mentioned, if 
you are willing to process the code a little you can count the characters 
of the longest line, and scale the text to match the final output. 
 A couple of points dropped on a font stil are readable, and many 
sentences still fit.


Another is to encode the whole thing, in other words convert all 
the "<" for example to escape sequences.  Now HTML will wrap everything 
automatically.

Your other alternative is to use the CSS  overflow propery
eg 
pre {overflow: auto} or overflow:scroll
The problem there is the inconsistency between browsers......

Good ones will add scroll bars only when needed. Common and bad ones 
will add horizontal and vertical scroll bars at all times for every 
pre box.  It's ugly.

You can specify overflow-y and overflow-x separately if you prefer.

Another option is to show in a textarea and set the rows & columns

one question re CSS. I have small template, where I put two images 
one under the other. You can look at http://www.xidys.com/hanka. 
But I will need to cut some images, and that fact destroys aspect 
ratio for me :-) I would like to ask, if, in CSS, I can define image 
the way that it would not be scaled? My definition looks like:

.photo img {
    width: 614px;
    height: 460px;
}

and in html:


<div class="photo"><!--[obrazek 1]--><img src="obrazky/IMG_1361.JPG"/></div>

I would like to have something like a face, where you insert image 
top-left corner, but it uses aspect effect, that once it reaches 
particular width or height, it is drawn to the face, but the other 
axis stays as-is, so the aspect radio is kept correct. With above 
definition, it simply scales image to defined width-height, and I 
have to properly cut image in xnview, counting pixel ratios .... 
and that is very boring job :-)

Petr: give the text container a fixed width, then set both left and 
right margins to "auto". this should center it.

yes. sadly lost a lot stuff when the previous host shut down in the 
same week as my hard-drive died. I had to retrieve what I could via 
the wayback machine, and too few CD backups.

I saved page locally, and have still ´ I suppose it's fault 
 of WYSIWYG software made with english people in mind only (or not 
properly configured).

No need to encoding special chars this way if You have characters 
in declared encoding (that's what <meta http-equiv="Content-Type" 
for).

Due to a very slow Internet connection, I need to make the FTP module 
of my website builder script more efficient so I don't send files 
unnecessarily. What I have in mind is:


1. Delete all the files in the website directory on my harddrive 
to eliminate all unused files.
2. Build the website to the website directory on my harddrive.

3. Download a list of the file names and creation dates from the 
website (all are in one directory).

4. Read the list of file names and creation dates from the directory 
on my harddrive (all are in the one directory mentioned in 2 above).

5. If a file is on the hard drive but not on the server, send it 
to the server.

6. If a file is on the server but not on the harddrive, delete the 
file on the server.

7. If a file on the harddrive is newer than a file on the server, 
send it to the server.


Has anyone already done this? Am I forgetting anything? Any pointers 
on how to do this?

Sorry about the CRLF ..... you don't want to be checking the timestamps 
on the server with a slow connection. Just hold the last updated 
value locally and if it changes then transfer the file.  Same for 
deleting ... else you spend all of your time checking on the server 
over a slow connection.    You could check the timestamps or hash 
the local value ... then if  the hash value of the source changes, 
transfer the updated version.     There's some code to do some of 
this in build-sie.r  http://www.rebol.org/cgi-bin/cgiwrap/rebol/view-script.r?script=build-site.r
 but it's a rebol-ish task.

if you are on an unreliable or slow connection, you might experience 
timeouts which will in turn cause network errors. I recently worked 
on a similar system and you have to basically wrap all code that 
access the internet in TRY and do a lot of error trapping and possibly 
some retrying to ensure that uploads and downloads of entire filesets 
are done correctly. The code in build-site.r will not do that, so 
you have to restart the upload if it fails.

Thanks, Henrik and Graham.

I found various FTP servers report dates differently. Also the dates 
may not include the timezone, so you would have to assume it is in 
the timezone of the server and get the timezone from the server another 
way. Because there are so many variants of FTP servers you would 
have to do a lot of research to make this reliable, and then you 
wouldn't be 100% sure it would not fall over with some obscure FTP 
server.

Louis -- a couple of pointers about uploading files to a server using 
a slow FTP connection:

(I do it myself with REBOL.org -- most of the development takes place 
on my machine and is uploaded to RO via a 56K modem, so this is based 
on real experience.)

-- If you are uploading a large live file, that file will be available 
and/or "broken" during the course of the upload. Best to upload with 
a temporary file name, and then rename when uploaded.

-- That won't work with CGI scripts under Apache/UNIX as the rename 
won't leave them with the right file permissions to execute. But 
it will work for all other files, including scripts that are DOne 
by your CGIs.

-- We have a checksums file that the uploader uses.  Before uploading 
a file, it checks the file's upload checksum. That way, we only ever 
upload new or changed files.

Thanks, Graham, for the script. And thanks Henrik, Anton, and Sunanda 
for the pointers. I'm hoping to be able to start working on this 
within the next few days.

This is very good CSS tutorial, which someone may find useful - http://www.thenoodleincident.com/tutorials/box_lesson/boxes.html

And this is another good place from which I found the link above 
- http://www.mandarindesign.com/

I had faced with problem for file uploads:

I had file upload script (upload.r posted somwhere in rebol lists/worlds) 
on my web servers, that works well on KF web server.

But on MS IIS script hangs and I am getting timeout error from server.
If it is needed I can post upload.r here!

Actualy problem is in function read-post data - script hangs on read-io.
Why it is problem for MS IIS and how to solve them? 

read-post-data: func [
    {Reads the HTTP entity body}
    /safe "Disables evaluation of content-length header."
    /local len data tmp
] [

    len: load any [ all [safe "65536"] system/options/cgi/content-length 
    "0" ]

    data: make string! len
    tmp: make string! len
    while [ 0 < read-io system/ports/input tmp len ] [
        insert tail data tmp
        clear tmp
    ]

    data
]

well, working with cookies is not all that difficult, is it? My friend 
just asked me - why rebol does not handle sessions, if any other 
language does. I told him to write it himself, but he probably does 
not know how. Isn't session just about getting a cookie, looking 
into your storage space for the cookie identifier (session identifier), 
loading the session data, using them, and storing them once again?

how should I design my function, if I would like to have e.g. session: 
copy [] block, and later would like to append whatever rebol value 
into it? e.g. variable names using in script, objects, etc? when 
I do append session var, it stores its value .... what would be the 
best aproach?

the simplest way, which however needs write permissions to the filesystem, 
is to have a unique session id assigned to users; this id could be 
basically a file name (and you need to check for its sanity then); 
then you read from the file at the beginning, and save to it at the 
end.

I just thought about how to store and later load some rebol values, 
add new values to them .....

temple.cgi uses cookies. there is no problem in handling multiple 
cookies on the server side, and the bug with multiple cookies on 
the client side has long been fixed.

maybe a pair of word name (literal) plus value ... and then some 
little accessor function

in some cases you may just need a logic! value, in others a block 
with words and values may be best...

ok, so you know you want to save variables: name last-name user-object 
some-block-here .... how do you save them, and later invoke them?

I was really tired by Bobik, who leaving rebol just slanders it, 
without proper deep knowledge of what session actually is. He reported 
to various other friends, that Rebol is terrible, because it is the 
only language not actually supporting sessions. Yesterday I looked 
into what "sesssions" mean in php terms, and I have to laugh - for 
me it is matter of few hours coding at max, to get it working ...

while he is negative, the positive thing for me is, that his opinion 
means, he likes compact design of Rebol in fact, and regards it being 
a platform in itself :-)

and we know that some people never get it sufficiently to be proficient 
enough

graham, yes many people try to use a new language like they've always 
done before... in any C derivative that works... cause basically 
you aren't doing anything differerently just writting it differently 
(java, perl, python)  but in others, you have to adapt and well... 
it doesn work!

I know - but then some ppl would have to understand what is behind 
the sessions - and there really is not magic :-)

I've just added a script to REBOL.org that offers session support 
in a CGI environment:

http://www.rebol.org/cgi-bin/cgiwrap/rebol/view-script.r?script=acgiss.r

Documentation: http://www.rebol.org/cgi-bin/cgiwrap/rebol/documentation.r?script=acgiss.r

And, yes, Petr, it really isn'y magic......Though I hope it is debugged. 
Please let me know otherwise.

Thanks C. It is really easy isn't it?

The main differences between your approach and mine that I can see 
are:
1. you hold all data in one file; mine uses one file per session

2. you create just a cookie; mine creates a session record in which 
the CGI script can save data

Either way, the scripts demonstrate that the problem is really trivial 
-- with the one need to create unique and hard-to-guess session ids. 
 We've both put some serious code into doing that.

Hi. Is it possible Google does not index properly our http://www.jablunkovsko.cz
site, because links to particular subsites are burried in java-script 
menu? I created account with google, and checked my robots.txt allows 
engine to visit my site ....

Javascript only links will stop pretty much every spider.....So you 
won't appear in many search engines at all.

You will also be breaking a lot of people's assistive technologies 
(like screen readers for visually disabled people).

Best to have a site that is fully navigable via plain HTML links. 
Then add the clever stuff on top of that.
Google's advice:

http://www.google.com/support/webmasters/bin/answer.py?answer=35769
http://www.google.com/support/webmasters/
***

A high score here usually means you have  a person-friendly and spider 
friendly website:
http://www.cynthiasays.com/

Is building and submitting a purpose built crawler page an outdated 
technique?

Search engines like to index things a user can see.

Thye get suspicious of anything that appears to be built just for 
the.....No surprising given how much they have been gamed by SERPS 
spammers.

Having a human-readable site map has the same effect as a crawler 
page *and* taks you a long way towards having a site that is navigable 
by people under any conditions.
***

The google-approved method of doing a crawler page is to use what 
they call a sitemap:
http://www.google.com/support/webmasters/bin/topic.py?topic=8476

Someone could easily modify Carl's site-checker to emit a google 
sitemap:

http://www.rebol.org/cgi-bin/cgiwrap/rebol/view-script.r?script=site-check.r

if there is nothing, you could id  the page and block further  replies 
 with the same id.

i guess i am clueless and beter  stop spamming. what is etags?^^

Etags should work like this: client requests a resource, server returns 
resource and etag header.  Client again requests resource, but this 
time sends if-none-match header with etag value and server can then 
decide whether to process the request or send 304 not changed.  I 
haven't explored this much so don't know how well it's supported 
-- I'm not sure how appropriate the solution is, but would involve 
the least server activity.

the problem is, that if you are posting large file, it's not too 
good, as all file is first uploaded again and than you can decide, 
if you want it to store for example. I though I sow something, but 
maybe it was just a fata morgana, I thing, that the best is the redirect. 
My problem is, that I'm trying to connect on database, which was 
not designed by me, so there is no crypted id for the data I upload, 
just incremental integer, which I don't want to use, as I would give 
chance to everybody go thru the data just inserting other numbers. 
I have to take a risk, modify the original database adding ne column 
with crypted id and redirect to the uploaded data using this id. 
And hope, that I will find and modifie all places, where is this 
database used:(

charles, welcome to reality !  In theory rebol supports many protocols 
but in practice the support is incomplete and it has been incomplete 
for a long time. check rebol.org, there are several patches for cookies 
support

Yes, I did, and if you scroll a little bit up, you can find the link 
easilly - it's in big yellow block of text:) And I agree, that it 
would be good to have cookies support directly in Rebol, as my cookies-daemon 
is relly hard hack I cannot be sure that it would not rewrite some 
future http protocol updates

I want to do some small post-card system for my friend. He can't 
program much and adapted small php solution, but that solution is 
so basic, that it even does not randomize postcard number in presented 
link ... so you can cange your url and get access to someone's else 
message. That is imo unacceptable.

I thought datatypes should be usefull, and there is exactly precise 
description of what is, and what is not email, so how Rebol can simplify 
it so much?

Not quite an email address validator, but the REBOL.org uses an email 
address spotter [contributed by Andrew Martin] to find and obsfucate 
email addresses in ML messages, etc.

You could use the same basic parse patterns to identify / validate 
email addresses. 

I haven't checked RFC 822 in detail, so it is likely that Andrew's 
code is more liberal than the actual spec, but it could be a starter.
Would you like a copy of the function?

Note to Sunada ...  licence (english) and license (american) spellings 
appear on the same page :)

I (finally) got around to fixing the license/licence typos. Thanks 
again Graham.

If you spot any other typos on REBOL.org, please report them in the 
Library group.
***

Incidently, I tend to use US english for webpages and British english 
for program code (variables, comments etc), so my code is littered 
with duplicate spellings: color/colour etc,

Hi,

I need another short advice :-)

do you prefer:


- directly accessible menu menu: http://www.creactive.cz/work/xidys/online/

or


- initially collapsed menu: http://www.creactive.cz/work/xidys/xidys_web_hp.jpg

   - with mouse-over effect: http://www.creactive.cz/work/xidys/xidys_web_hp2.jpg

   - and after clicking an item: http://www.creactive.cz/work/xidys/xidys_web_sec.jpg

(dis)advantages to both:


According to company working on our website, there is something like 
slight barrier, when user needs to click, and wait where the click 
does take him/her. According to them, users prefer scrolling. OTOH 
the menu can get long, and e.g. e-shop direct icon is not visible 
(nor is last section), and there is no easy other place, where to 
put such an icon (direct link).

thanks a lot :-)
-pekr-

I definitely prefer all the info being visible on the page.

Expecting people to play hide and seek is good in games and some 
styles of websites. But if you are trying to make your site (and 
therefore your company's information) easily available, then don't 
hide links in any way.

ok, so far that seems to be option of circa 60% of ppl I asked :-) 
40% prefer partially collapsed aproach, and noone prefers pop-up 
menus :-)

Yes, but then menu is too deep to be visible - and what in menu is, 
and what is not, - crucial :-)

It's trade-offs Petr!

Dropdown/pullout menus take less screen space but need more clicks, 
and may be less accessible.

Too many visible options on one screen make it difficult too see 
the important links.
Its not easy to get the balance right on the first try.

That's why usability testing is needed, and why usability consultants 
can make a good living.

Pekr, I prefer the expanded menu, immediately available. I don't 
care if it's two or three pages long. If the section titles are bolded 
(eg. "Internet a site", "Kiosky" etc.) and obvious, then reading 
them while scrolling down the page is no problem. And I think your 
search bar should go at the top, to solve the problem of being too 
far down to be visible initially.

simply put, there is following workflow, similar to how Gabriele's 
Temple worked:

- browser requests page
- webservers takes .html
- then it sees, .html should go via some handler
- handler is called - so e.g. index.html is passed to pekr-rsp.r

- pekr-rsp.r, to save some processing time (prevent parsing), looks 
if .html page is registered as dynamic, non-dynamic .html is returned 
directly to browser - non-measurable slow-down

- if page is supposed to be dynamic, it is being parsed. It looks 
for marked sections and particular handlers are called, which know, 
how to treat that section

remark MOD will allow all you are saying above  :-)  that's the point. 
 the advantage is that you can hide and integrate all those tricks 
on a programmable per site/ per page / per element .  since each 
tag receives the content its handling, if your outer tag can handle 
caching on the server side, well just do so.


remark changes HOW you can do your tags cause they are NESTED.  and 
can return dynamic stuff... not just html.

well, that will depend on what you mean by template and what your 
template contains... if you talk about a frame (or various sub parts 
of a page) which hold html, yes remark can have these with no remark 
tags within... so you can linkup an html page based on artists work, 
and add up your dynamically created content.  But remark even allows 
you to programatically include those little html parts at any level, 
so one of your dynamic tags could in fact be loading just menu titles 
which which your gfx artist created manually.  but their placement 
or the choice of which set to load would be controled within the 
dynamic tags.

max - simply put - what will any <cahe!> tag, which is NOT even paired, 
do, if you add it to html source, and you display it?

the cache! tag would be a wrapper you add to some "constructor" page.. 
the actual elements which compose your page, are still visible and 
useable directly.

I would like to resolve following situation:


- on our kiosk, we have following set-up - Apache running locally, 
and once Windows starts, it runs http://localhostin full screen.

- I wanted operator to have more easy life (not taking wireless keyboard 
each time), so I wrote View script, which sits in the memory, and 
waits for USB drive to appear. Then it copies new stuff from USB 
drive to C:\kiosk dir.


However - we found out, that the presentation still returns old pages. 
They seem to be cached somewhere. What should I do about it? Improve 
script to:

- restart presentation?
- restart Apache service?
- disable caching somehow?
- flush the IE cache somehow? (could take long)


And remember - the aim is for the operator to have seamless operation 
- plug-in USB disc, wait for update, unplug disk.

Guys I have a short question: Is there a simple and fast way to avoid 
that a html-formular is position back to the top after a drop-down 
box has been selected which triggers a server call?

Any DOM experts here ?

I have a frameset with two frames, and in the first frame there is 
a menu implemented with <ul>.

I am wondering if I can move the <ul> menu out of the frame into 
the top level document. (The frame exists only for the menu, so I 
should be able to size it to zero so that it does not obscure the 
top-level doc.)

This would mean I could keep the website frameset-oriented without 
doing the work of converting it to single pages. And frames allow 
pure HTML to include code efficiently.

I found document.body.removeChild(elem), but I get lost when I try 
to identify the new location and createElement.
Can anybody help with that ?

load the jquery library into your site and make your life much easier, 
it makes manipulating DOM a kid game 8) http://jquery.commaybe look 
also for the accordion or tree plugin for your menu

There's an approach that I'd like to experiment regarding web UI 
generation. The idea would be to consider the web page as a View 
target and build a View-like rendering engine able to process face 
objets and renders them as HTML. So it would be possible to generate 
HTML UI with VID directly. The DIV tag would be a good candidate 
to emulate a View face.

The events system would be splitted in two parts :


1) A small part in JS on client-side catching keyboards and mouse-events 
and sending them asynchronously to the server.


2) A server-side part emulating the View event propagation engine 
through the face objects hierarchy.

as for JS, I am not sure it is fast enough. I saw Sun's JS based 
desktop, and it was joke compared to even VID 1 alpha speed probably 
:-)

Keyboard and mouse events happen too quickly, and people expect them 
to be handled quickly.

With the DOM for regular UI elements, and Canvas for Draw once it 
is implemented more widely. The browser is a little weak right now, 
but that is improving, finally.

View-like engine in JS : that looks very doable, but I'm afraid the 
result might be too slow. I've worked some time with a rendering 
engine built in JS, and it was just a little slower than native UI, 
but not very scalable (performances dropping rapidly with a growing 
number of "faces"). It was BackBase : http://www.backbase.com.

So maybe the solution would be to handle those events that need fast 
response like typing text in a field, or mouse-over gfx effects, 
on the client-side and send others to the server.

If we are going to use a JS framework, I insist in using jQuery, 
the core is really mature,fast and stable, it has the syntax that 
more resemble the way we think in rebol. http://docs.jquery.com/Main_Page

Brian, no, Qtask hasn't, and that's what i've always been fighitng 
with Reichart about :)

would be nice if you can take control and update on rebol.org to 
prevent others headaces 8)

maybe it is just hidden, using cookies? If you have cookie, you are 
auto-logged-on, and the session will end when you close your browser 
session?

Web apps are becoming sophisticated. We are about to introduce WCMS 
in our company, probably basing it upon SharePoint Server. But I 
can see that also many open-source systems exist, as Joomla, Drupal, 
Xoop, WordPress, etc., and here's how you can customise your google 
site - http://www.google.com/ig- you simply choose portlets and 
you can shift thema around ....

It's up on REBOL.org, and I sent it to Doug Crockford as well.

Info from danish news on tv:
Women use 29 hours on the internet per week.
Men use 27 hours on the internet per week.

People older than 61 years use more time on the internet than people 
between 20 and 60 years old.

You might have to do some window spying to see what classes things 
are, then enum top windows to find the browser, then enum its child 
windows and do getwindowtext.

here is a more complete example that will save url of all browser 
windows to a file:
http://daringfireball.net/2003/02/save_and_restore_safari_urls

Brock,  yes dynamic content is hard to track and it depends if it 
makes sense to track.

this is part of a script I use to monitor and restart  my vpn and 
ssh tunnel :
connect: does [
		call rejoin [{osascript -e "tell application \"System Events\"
			tell current location of network preferences
				set vpn to current configuration of service \"VPN\"
				if not connected of vpn then
					tell application \"AlmostVPNProMenuBar\"
						stop (profile named \"default\")
					end tell
					connect vpn
					delay 6
					tell application \"AlmostVPNProMenuBar\"
						start (profile named \"default\")
					end tell
				else
					tell application \"AlmostVPNProMenuBar\"
						stop (profile named \"default\")
					end tell
				   disconnect vpn
					delay 6
					connect vpn
					delay 6
					tell application \"AlmostVPNProMenuBar\"
						start (profile named \"default\")
					end tell
				end if
			end tell
		end tell"}]
	]

My friend who is building some basic websites was asked by his client 
for the search facility. So far he used Google and its: site:your-domain.com 
feature, but the client would like the output to be customised. So 
here goes the questions:


- does google licence allow customising such output? E.g. wrapping 
it in another color, etc., so that it would look like the rest of 
the client site?

- is it easily technically possible, to parse the result? I mean 
- I can imagine parsing the output, even page paginator, then generating 
custom output. But maybe there is some other solution for that? Maybe 
some google api?

I think I will try CGI and parse ... hopefully I will not break some 
license, albeit I fear that I will be :-)

Oh, and it should run on the web-server.

Hi Robert  I switched from analog to awstats  about 4 years ago and 
am still with it.

Only two hosts returned #[none]:
	www.ted.com
	events.ccc.de
and the rest returned strings of lengths, eg: "4279194".

I should test more urls to feel more certain. (Or read http specs... 
hmm.. which gives better results ?)

Steve, could you be a little more specific.  There are a lot ways 
to parse your parse question :)


For example, are you asking what the "?" and the "&" mean in the 
URL?  Or am I taking too literal?