From f00025a39bf51f58c65a2c213e5b4a8741f40416 Mon Sep 17 00:00:00 2001 From: Jonas Wielicki Date: Fri, 20 Apr 2018 12:48:44 +0200 Subject: Explain how we speak before the server --- README.rst | 1 + 1 file changed, 1 insertion(+) diff --git a/README.rst b/README.rst index 17eea4a..c0d72bf 100644 --- a/README.rst +++ b/README.rst @@ -11,6 +11,7 @@ Tricks and shortcuts * We use ``tr`` to convert ``>`` to ``\n`` -- since sed is line (or NUL) based, there’s not really another way to parse XMPP XML (which generally never contains newlines) with sed. * TLS is handled outside of sed for similar reasons. And to keep my sanity (some people might question whether I still have any bit of sanity left). * Likewise, SRV lookup and composition of the authentication data is entirely handled in bash. This also means that only PLAIN SASL authentication is supported -- SCRAM requires a level of interactivity which would be extremely hard to achieve in sed (not impossible though; we would "just" have to implement base64 and sha1-hmac in sed). +* Since XMPP is a protocol where the client speaks first, we need to hand sed some initial input to allow it to generate a "line" of output (the stream header). We do that with bash, and use that opportunity to pass some configuration to the sed program (namely JID and authentication string). Usage ----- -- cgit v1.2.3