Skip to content

HTTP 1, 2, and 3 #91

Open
Open
@philsturgeon

Description

@philsturgeon

Braindump found in some notes somewhere, not entirely sure I know wtf I'm getting at here but here it is until I have something better. Chip in folks.


Http 0.9 pretty much just static docs
Http 1.0 what did it improve I guess Ajax now
Http 1.1 keep alive 10% improvement, lots of clients don’t use it, Wework graph, the gold standard until spa meant lots of connections, browser limitations
Lack of knowledge about keep alive has lead to folks batching requests to avoid “handshakes” which is often actually slower

Let’s stop building to take into to http2, leave model design until later

H2 allows multiplexing so lots of requests made over same connection and responses are mixed together, browser solves max connections from hard set 6- to confugurable 100

H2 hpack means headers are squashed (handy if you have big JWT and other session data, but try to avoid that anyway), and server push is awesome. Please push mention.

Http 0.9 going away

curl/curl#4191

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions