This allows the input callback function for a buffer to receive multiple
lines at once, instead of the message being split on newline before
being sent to the callback. It adds a new flag, input_multiline, to
control this. This flag defaults to 0 which is the current behavior.
New comparison operators:
- "==*": is matching mask, case sensitive (wildcard "*" is allowed)
- "!!*": is NOT matching mask, case sensitive (wildcard "*" is allowed)
- "==-": is included, case sensitive
- "!!-": is NOT included, case sensitive
- "=-": is included, case insensitive
- "!-": is NOT included, case insensitive
This introduces a new command called "handshake" in the weechat relay protocol.
It should be sent by the client before the "init" command, to negotiate the way
to authenticate with a password.
3 new options are added:
* relay.network.auth_password
* relay.network.hash_iterations
* relay.network.nonce_size
Colors changed:
* relay.color.status_active: lightblue -> green
* relay.color.status_auth_failed: lightred -> lightmagenta
* relay.color.status_connecting: yellow -> white
* relay.color.status_waiting_auth: brown -> yellow
The color relay.color.status_disconnected remains lightred by default.
Now the status "waiting_auth" is used and displayed in the relay buffer.
When a client connects, there are now 2 messages (except for irc protocol if
there's no password required):
relay: new client on port 9000: 1/weechat/1.2.3.4 (waiting auth)
relay: client 1/weechat/1.2.3.4 authenticated
If the authentication fails, the messages are:
relay: new client on port 9000: 1/weechat/1.2.3.4 (waiting auth)
=!= relay: authentication failed with client 1/weechat/1.2.3.4
The memory leak was caused by a bug in function setlocale on FreeBSD:
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=243195
The fix is the following:
* Remove the calls to setlocale when formatting the result.
* The function snprintf is still called, and then is now locale dependent,
for example in French the decimal separator is a comma instead of a dot.
* A new function calc_sanitize_decimal_number is introduced to "sanitize" a
decimal number: keep only the decimal separator (replace it by a dot) and
remove any other separator found.
Unit tests are added on these functions:
* calc_sanitize_decimal_number
* calc_format_result
The evaluation allows to use different sort for the bar items, for example with
such value:
${if:${bar_item.name}==buflist?number:short_name}
This sorts by number on the first bar item and by short name on the two other
bar items.
