You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The devel+sharedbuffers branch is a prototype of this; IIRC the speedup on the chanflood benchmark was about 30%. We'll probably get a bit less.
Here's the proposal: for each PRIVMSG/NOTICE/TAGMSG, cache 2 versions of the built message. One is for clients with message-tags negotiated: these clients are eligible to receive account-tag (where applicable), server-time, etc., so we'll send them a cached "full" message with all tags. The other is for clients that have no tag capabilities negotiated (neither message-tags, nor server-time, nor account-tag, nor anything else that would indicate the ability to receive tags): they'll get a cached message with no tags. Other clients will go through the current flow where we build an individual copy of the message.
The text was updated successfully, but these errors were encountered:
The
devel+sharedbuffers
branch is a prototype of this; IIRC the speedup on the chanflood benchmark was about 30%. We'll probably get a bit less.Here's the proposal: for each PRIVMSG/NOTICE/TAGMSG, cache 2 versions of the built message. One is for clients with
message-tags
negotiated: these clients are eligible to receive account-tag (where applicable), server-time, etc., so we'll send them a cached "full" message with all tags. The other is for clients that have no tag capabilities negotiated (neither message-tags, nor server-time, nor account-tag, nor anything else that would indicate the ability to receive tags): they'll get a cached message with no tags. Other clients will go through the current flow where we build an individual copy of the message.The text was updated successfully, but these errors were encountered: