Willy Tarreau 0e9b1b4d1f MEDIUM: compression: consider the "q=" attribute in Accept-Encoding ...

Till now we didn't consider "q=". It's problematic because the first
effect is that compression tokens were not even matched if it was
present.

It is important to parse it correctly because we still want to allow
a user-agent to send "q=0" to explicitly disable a compressor, or to
specify its preferences.

Now, q-values are respected in order of precedence, and when several
q-values are equal, the first occurrence is used.

2014-03-19 12:12:01 +01:00

..

acl.c

MEDIUM: acl: Change the acl register struct

2014-03-17 18:06:08 +01:00

appsession.c

…

arg.c

…

auth.c

…

backend.c

…

base64.c

…

buffer.c

…

cfgparse.c

…

channel.c

…

checks.c

…

chunk.c

…

compression.c

…

connection.c

…

cttproxy.c

…

dumpstats.c

…

ev_epoll.c

…

ev_kqueue.c

…

ev_poll.c

…

ev_select.c

…

fd.c

…

freq_ctr.c

…

frontend.c

…

haproxy-systemd-wrapper.c

…

haproxy.c

…

hash.c

…

hdr_idx.c

…

i386-linux-vsys.c

…

lb_chash.c

…

lb_fas.c

…

lb_fwlc.c

…

lb_fwrr.c

…

lb_map.c

…

listener.c

…

log.c

…

map.c

…

memory.c

…

pattern.c

BUG/MEDIUM: acl: boolean only matches were broken by recent changes

2014-03-17 20:12:40 +01:00

payload.c

MEDIUM: acl: Change the acl register struct

2014-03-17 18:06:08 +01:00

peers.c

…

pipe.c

…

proto_http.c

MEDIUM: compression: consider the "q=" attribute in Accept-Encoding

2014-03-19 12:12:01 +01:00

proto_tcp.c

…

proto_uxst.c

…

protocol.c

…

proxy.c

…

queue.c

…

raw_sock.c

…

rbtree.c

…

regex.c

…

sample.c

…

server.c

…

session.c

…

sessionhash.c

…

shctx.c

…

signal.c

…

ssl_sock.c

MEDIUM: acl: Change the acl register struct

2014-03-17 18:06:08 +01:00

standard.c

…

stick_table.c

…

stream_interface.c

…

task.c

…

time.c

…

trace.c

…

uri_auth.c

…