Le Web est mort, vive le Web!

A bit of historic perspective this morning...

John Gruber has posted a thoughtful discussion of Google's new Chrome OS to his Daring Fireball blog, in which he cited the VT-100 as an example of a popular client-server system. Unfortunately, his usually accurate and insightful commentary is incorrect in that detail today.

A client-server system is one in which a client, which has a degree of intelligence and computational ability of its own, requests data from a server, then manipulates it locally.

The VAX microcomputer terminals, by contrast, provided a nice interactive terminal, but they weren't really client server — just dumb terminals with cursor repositioning. A big step up over mainframe screen-based systems, but not really client-server. (See my 2006 Down with the Web blog for details.)

VT terminals were just remote Keyboard-Video setups for multiuser computers. Even the graphical X-terminals DEC later introduced had no intelligence beyond the screen drawing and mouse system. When the VAX was introduced, client-server wasn't even a concept yet.

What's that got to do with the web?

Again, let me refer back to my three-plus year old Down with the Web rant, in which I declared that web browsers were nothing more than glorified mainframe terminals.

Things were starting to change at the time, but in the ensuing three years, they've changed a lot. Many "web apps" can be thought of as modern lightweight client-server applications. For example, the Gmail web interface on iPhone loads an application into Safari; that app then requests data from the Gmail servers, parses them out, creates a UI and presents the messages for action. It can even make some calls to other applications on the iPhone, such as the phone app, YouTube and Google Maps. Classic client server setup — grab the server side data, do smart things locally. Similarly, Yahoo! Mail for years has been a full-on web application.

While PC-based client-server systems of the early- to mid-90s stored the code locally, web apps store their code on the server. So like the ill-fated "network computer" Oracle was promoting in the late nineties and early aughties, they're network dependent... but no less client-server in their fundamental architecture.

Better late...?

So, a few years later, I have my answer. The web's antiquated mainframe architecture is slowly on its way out, as it turns into a true client-server application platform.