karlcow

Opening The Web one bug at a time

HTTP versioning

Wednesday, June 29, 2011 6:53:04 PM

, , , ,

HTTPbis WG is wondering if they should change the version number of HTTP and in which ways. A bug has been opened to define the version number as a fixed length pair of Digit. Willy Tarreau mentionned that

Too many implementations I’ve seen just check for ver[7] == ‘1’ or just memcmp(ver, “HTTP/1.1”, 8). If we ever decide to use 1.10 as a valid value, we’ll get a lot of trouble due to that.

It reminds me of the hell of Opera versioning

But in my opinion, the real questions are:

  • What are the most common libraries with different behavior/code path depending on the HTTP version number
  • How many of these libraries will break if the version number is misinterpreted?

Velocity 2010 - Take it all off! Lossy Image OptimizationTwo APIs For Address

Comments

Charles SchlossChas4 Tuesday, August 9, 2011 9:29:22 PM

Google should be having the same problem as Opera with Chrome since they are on version 13.x

Karl Dubostkarlcow Wednesday, August 10, 2011 12:37:35 PM

Yup they do, with a slight difference. Developers are testing in these browsers and so developers notice the mistakes quickly.

Charles SchlossChas4 Wednesday, August 10, 2011 3:19:15 PM

Tho it is because Opera was first

Originally posted by karlcow:

so developers notice the mistakes quickly.


that they can't count past 9 bigsmile

Write a comment

New comments have been disabled for this post.