https://bz.apache.org/bugzilla/show_bug.cgi?id=63835
--- Comment #11 from Mark Thomas <ma...@apache.org> --- Mainly because I want to get 9.0.x and 8.5.x tagged ASAP I am intending to commit fixes to address the concerns raised here shortly. I agree a rename of the constants would help. My reading of the code and RFC 7230 is that it is acceptable to send this header to HTTP/1.0 clients when this code does this. The short form of my reasoning is: - keepAlive is set to false for HTTP/1.0 requests - only if Connection: keep-alive is present is keepAlive set to true (i.e. the HTTP/1.0 client has explicitly advertised keep-alive support) - this header is only sent if keepAlive is true I agree that this new feature should be configurable on the Connector, enabled by default. I thought I had stated somewhere I thought this should be configurable but I can't find a reference to that at the moment. Making it configurable also addresses any issues with HTTP/1.0 if the reasoning above is wrong. I agree the behaviour is "odd" for timeouts < 1000ms. Given the spec doesn't cover this, consistency with httpd (i.e. unchanged from current) as as good as option as any. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
