Eureka! (Re: Comments on Python Redesign)

Mon Sep 8 19:07:23 EDT 2003

On Mon, 08 Sep 2003 17:13:58 -0400, Peter Hansen <peter at engcorp.com> wrote:

>Skip Montanaro wrote:
>> 
>>     Dave> I don't know of any problem with tinyurls - *any* server can
>>     Dave> provide a "bad" redirect; as with any URL whether or not you
>>     Dave> follow it depends on the context and on how much you trust the
>>     Dave> sender.
>> 
>> The only other reasons I could think of to object to tiny urls are:
>> 
>> * Perhaps the tinyurl folks are not trustworthy (are they keeping track of
>>   your browsing habits and selling that info?).  That seems unlikely, since
>>   they don't appear to be storing cookies in my browser.
>> 
>> * The tinyurl website can become a bottleneck, preventing people from
>>   getting where they want to go.  The tinyurl.com website seems to be fairly
>>   unreachable for me at the moment, though this is the first time I've seen
>>   this, and it appears to be a problem more with Northwestern's connection
>>   to the net than tinyurl.com's.
>> 
>> * There's no guarantee tiny urls will remain unique or even available for
>>   long periods of time.  Their current practice of using four lower-case
>>   letters suggests they have space for about 450,000 unique URLs before
>>   extending to a fifth letter.
>
>* You can't tell ahead of time what site you are going to be redirected to.
>  As a result, you lose complete freedom to choose whether or not you should
>  follow the link.  (Note for the imprecise: I didn't say you completely lose
>  freedom, I said you lose complete freedom.)
>
I guess it wouldn't be too hard to set up a cgi url on python.org, if desired,
to do two-step redirection -- i.e., you'd pass it a small string and it would look
up the URL and generate a minimal html page with a plain url link in it, e.g.,

<html><head><title>www.python.org coded url service</title></title></head><body>
   <a href="http://the.real.target.com">http://the.real.target.com</a>
</body></html>

And then you could choose whether to click, but you only have to type e.g.,

  http://codedurl.python.org?smallstring

or such.

And you could accept URLs via a form and generate and store the short codes. Some filtering
would probably be advisable.

As far as tinyurl itself, I wonder if one couldn't get the real url (without committing to go
there) by doing the http protocol step by step with telnetlib.

I'm not feeling motivated enough to do it at the moment ;-)

Regards,
Bengt Richter