What did www.python.org look like from 1996 to 2021?

I started reading the first snapshot they had (1996), and noticed that there was once a web browser called Grail which aimed to be a web browser completely written in Python: https://arquivo.pt/wayback/19961013225840/http://monty.cnri.... Very neat stuff. Almost 30 years later, I bet someone would have enjoyed that. (If it had taken off, we might be scripting web pages in Python instead of Javascript, even!)

The website is actually still up: http://grail.sourceforge.net/ The latest minor release is from 1999.

At least edit the video to remove the archive redirs. Pretty poor job that.

-1 for hijacking the browser’s back button

All these links are prefixed with https://href.li which uses meta refresh as a means of redirection, which apparently still works on most popular browsers, to "hide" the Referer header. What I have never understood is why no one cares if the operator of href.li thereby gets logs of every URL submitted along with originating IP. In this example, there is an implicit assumption that the user wishes to "hide" the Referer header from one website, e.g., arquivo.pt, archive.org, bibalex.org, vefsafn.is, loc.gov, webcitation.org, or archive-it.org, but not from another, i.e., href.li.^1 This is, IMO, a recurring theme with "free web-based service"^2 websites and "tech" companies. One website/"tech" company expects users to trust it but to distrust others.^3 Nevermind that the choice might sometimes be between the lesser of two/multiple evils, so to speak.

1. Though I suspect here it is the operator of mementoweb.org who has the interest in hiding the Referer header, for whatever reason.

2. The easiest way to avoid the annoyances of Referer headers is not to send them. Unless things have changed, sending a Referer header can be disabled in Firefox. A local forward proxy can also remove the Referer header. Smaller, open source HTTP clients that users can fully control and modify need not send Referer headers. With possible rare exceptions, only ever raised by website operators and "tech" employees, sending the Referer header serves no benefit to users. It should be optional, but the most popular browsers remove that choice.

3. Google is an obvious example. DuckDuckGo and other alternative search engines have also been known to prefix URLs to create redirects, purportedly to "protect" users form the dangers of the Referer header. This of course also allows the search engine to track every URL (search result) that a user "clicks".

I did something similar for kde.org a while ago, in case someone is interested https://carlschwan.eu/2020/04/18/a-new-look-for-kde.org/

I’d be surprised if they had HTTPS support in 1996.

I liked all the versions up to 2005 (the one with the blue sidebar).

If you put in a blocked site [1], interestingly a picture of j robert oppenheimer comes up telling you that this is a service from los alamos national laboratory.

Shame they blocked deoxy, that was one of the oldest websites on the internet. They even had a version from the early 90s which was meant to be browsed in a text-based browser.

[1] http://timetravel.mementoweb.org/list/20000705010912/http://...

Warning: unnecessary loud music accompanies the video.

Interesting. I like the 2006 aesthetic better, but I like how the current one puts code front and center.

The change I'd make is to add a big-ass "Download" button somewhere. That has to be the number one reason why people visit python.org, after all.

Current design looks too heavy on blue to me.

2004 is by far my favorite. Seems really clean and intuitive.

Why does this page prevent you from backing out in one click?

And now for R. (Only one snapshot needed.)

What's the music?