Hacker News Reader

Comment by denton-scratch | Hacker News Reader

> The layer that was—and still is—responsible for standardized text transfer on top of TCP/IP: HTTP

Um, wait. HTTP is for transferring documents. There are other text-based protocols that run over TCP, that long pre-date HTTP. And I for one deplore the use of HTTP for everything; I'm afraid the common firewall practice of blocking everything outbound except port 80 is probably largely responsible.

View on HN

HTTP represents any documents as ... drum roll ... text.

It is literally in the name of the specification.

If it were meant to be a document transfer protocol, HDTP would have been a better acronym.

If you can rely on the name of the protocol as an accurate description, then LDAP would actually be "lightweight", UDP packets would be assembled by users, and SSL would be secure.

If HTTP represents documents as text, then the "hypertext" part of the name becomes misleading, because you can't put hyperlinks in plain text. That is: the protocol is misnamed, because the leading "HT" stands for "Hypertext", not "Hyper Text". It's designed to transfer HTML.

> If you can rely on the name of the protocol as an accurate description, then LDAP would actually be "lightweight", UDP packets would be assembled by users, and SSL would be secure.

Also SOAP would be simple: http://harmful.cat-v.org/software/xml/soap/simple

Original HTTP (now called 0.9) did not have headers in the request. HTTP 1.0 added headers, including the "Content-Type" header (reused from the MIME e-mail standard), specifying the type of the data being transferred. Before that, all data was indeed implied to be HTML and nothing else.

Reply on news.ycombinator.com