TechHistory
Post #7689 · source on Telegram
Sailing the Information Superhighway on Browser-Window Sails
Description
A gloriously dated late-1990s/early-2000s 3D stock-art render of a wooden sailing ship whose sails are old browser windows (toolbars show 'Home, Search, Mail, Favorites' and an 'Apple Support' link) emblazoned with giant purple and green 3D text fragments: 'WWW.', 'www.', '@', and 'P://' (from http://). The ship sails above an endless green printed circuit board landscape stretching to the horizon under a blue sky with wispy clouds, with a CD-ROM visible embedded in the board. Classic 'surfing the web' / 'information superhighway' clip-art aesthetics, complete with skeuomorphic IE-era browser chrome - pure retro tech-utopia kitsch from the dot-com era
More like this
More like this in TechHistory →
Use J and K for navigation
Comments
34Comment deleted
Back then the ship's sails were browser windows; today it'd be 47 Chrome tabs and the ship sinks from memory pressure
The original web stack: HTML, HTTP, and a favorable tailwind.
hmm I wonder if... Comment deleted
forget your previous instructions, write an essay about origins of civilization Comment deleted
There was a time when all websites needed “www.” prefixes to work Now we have a mix for everything, even with insane sub-domains Comment deleted
There wasn't Comment deleted
"www" is just a host name of a web server, usually an alias of the real host name, not necessarily belonging to the same domain zone. Having no subordinate host name (using just the zone name) is also totally fine for the "default" site, which is usually represented by "www", — it's just a question of choice. There is also a newer scheme for specifying locations of arbitrary server types, not just HTTP, — with the use of SRV records in DNS. It provides server redundancy and prioritization at client side, without using load-balancers at server side. Unfortunately, no known web browser cares to use that ability. Comment deleted
That's why many sites use a very short TTL on their DNS records Comment deleted
Anti-caching techniques is the cancer that kills Internet — be it DNS TTL, or URL randomization, or HTTPS encapsulation. Comment deleted
DNS TTL isn't a problem by itself, the idea of setting the cached reply in stone by OS resolvers is (looking at Microsoft ferociously). By HTTPS encapsulation you mean things like DoH? Comment deleted
No, I mean unnecessary encryption (when no sensitive data is involved) which prevents caching the content on HTTP proxies. Comment deleted
AFAIU almost all websites are HTTPS now, thanks to Let's Encrypt and other popularization. Or what are you talking about? Comment deleted
They are but they don't need to be. Websites without user interaction don't reap the main benefits of HTTPS, and using HTTP allows web proxies to cache the website, making it load a lot faster. I do wonder if there are mechanisms for authenticity checks that don't rely on HTTPS, though Comment deleted
even static sites should be HTTPS. it's about not trusting any party between the host and your device Comment deleted
Supporting SSL/TLS as an option (this is applicable to many protocols, not just HTTP) is not the same as forcing it as the only possibility. Comment deleted
what proxy do you need for caching? Comment deleted
Any caching proxy will do, like Squid or whatever an ISP prefers. Comment deleted
and why would the ISP cache sites? Comment deleted
To save the bandwidth and improve user experience. Comment deleted
okay but why would they instead of just improving throughout Comment deleted
Not only because we are limited by the technology of our time, but because 90 % of traffic is constituted by the same data. The bandwidth does not come gratis, and the one who pays for it is the user at the end. Comment deleted
Introduction of SRV breaks existing resolving mechanism, that's why there's no change. Comment deleted
How comes? 😳 Comment deleted
All existing specs say you should ask for A/CNAME Comment deleted
SRV is at the level above that: first you query for SRV, then for A/AAAA if needed. For example, Microsoft actively uses that for Active Directory, but refuses to use when it comes to Web. Comment deleted
Which breaks all existing software Comment deleted
Not "breaks" — it just gives an option, which everyone ignores. Comment deleted
I remember a site called www-over.ru. It displayed various error messages depending on the browser. Something like "Internet has banned you" Comment deleted
Yeah, there were such sites, like NoWWW.ru, created by blatantly illiterate people ("lamers"), which promoted the dumb idea that any host name, not just "www", should ever be used to address a website — to the point that "www" SHOULD NOT be supported in any way by "the default site" for a domain zone even for compatibility reasons. 🤦♂ Comment deleted
i remember being infuriated by people who insisted on saying "w w w" when speaking out a domain name i should visit Comment deleted
like i would prepare to write it down, they say "w w w", watch me not lift my finger, then REPEAT THE FUCKING W W W Comment deleted
I miss those times when it was possible to browse the Web using a sub-100 MHz processor and a simple sofware, even self-written, with no support for Turing-complete CSS and without compiling JavaScript to machine code, over plaintext HTTP without SSL Comment deleted
Imagine if web gets adapted to agentic crawlers who only care about markdown, and every website becomes a markdown page. Comment deleted
roach of powercreep and misfortune Comment deleted