[rabbitmq-discuss] Cannot send message with STOMP

[Steve Powell]

Lionel,
Thank you for reporting this problem with our header en/decoding.

The STOMP 1.0 specification is unsatisfactory in many respects, not
least because it says nothing about encoding headers (or what it means
by CHAR in its BNF).

The STOMP 1.1 specification has drawbacks of its own -- the rather
arbitrary notion of escaping colons in header values as well as names
(why?) and the insistence of UTF-8 for all header names and values (this
is not policed in RabbitMQ since this would involve another (complex)
parsing process -- and it is not clear what we should do with non-UTF-8
octet streams). Finally, it doesn't correctly describe the escape rules
in respect of backslashes.

STOMP 1.1 also prescribes (in its improved BNF) OCTETs in headers and
header values (whereas the text (using MUST) insists they consist of
UTF-8 sequences) and it (the BNF) allows backslash (followed by anything
other than c or n) in a header.

So I think it is _incorrect_ in that it specifies a considerable
superset of what you can receive (or are allowed to send) over the wire.

When RabbitMQ started to support STOMP 1.1 we made it backward
compatible with 1.0 clients as far as possible. So we do not police some
of the verboten practices in 1.1 -- for example, we do not insist upon
UTF-8 strings in header names and values, and we still allow
subscriptions without ids. (Policing colons in header values (as well as
names) was probably a mistake in retrospect.)

We managed to produce a STOMP adapter implementation which avoids having
complicating checks of the connection version on frame parsing and
generation. It looks like this will have to change.

I'm going to raise a bug [24896] to track this, and propose the
following solution:

We use the negotiated connection type to determine the parsing rules
(and generating rules) for frame headers.

For 1.0 connections we will allow OCTET streams for header names and
values, excepting colons in names, and excepting newlines (\u000A
newline character coded in UTF-8 is x'0A') anywhere, and do NO escaping
either on input or output. We will reject no headers on input unless
they do not contain a colon before the first newline (because they
cannot be malformed under these rules) though the remaining frame
parsing can fail.

For 1.1 connections we will allow OCTET streams for header names and
values, excepting colons, newlines and backslashes, interpreting escape
sequences for these, and generating escapes for these characters on
output. We will NOT reject non-UTF-8 OCTETs but instead just pass them
through asis in order to allow the simplest client access. We will NOT
reject other escapes (\ followed by non c, n or \), but leave these
asis.

Although this is not strictly what the spec says (we would not enforce
UTF-8 or the errors on bad escape sequences) I believe it is a
reasonable compromise.

Steve Powell
steve at rabbitmq.com
[wrk: +44-2380-111-528] [mob: +44-7815-838-558]

On 17 Apr 2012, at 12:02, Tony Garnock-Jones wrote:

> On 17 April 2012 01:34, Lionel Cons <lionel.cons at cern.ch> wrote:
> IMHO, the BNF definition is correct: it accurately defines what you
> can see on the wire and this is sufficient to extract frames. It does
> not however define how to interpret what you extract from the wire,
> like backslash escaping and UTF-8 encoding, which are defined elsewhere
> in the STOMP 1.1 specification.
> 
> Fair enough. Perhaps "incorrect" was a bit strong, but it could definitely have helped this casual reader avoid the dangerous misinterpretation I ended up making by reading the BNF alone. I believe it could profitably also describe the backslash-escaping. (After all, it would also be correct, in some sense, for it to specify the headers block as "a sequence of bytes not including two newlines in a row, followed by two newlines", leaving it up the the rest of the text to describe how to extract the key-value pairs from that structure! ;-) )
> 
> Here's the kind of thing I was expecting:
> 
> header              = header-name ":" header-value
> header-name         = 1*header-text-atom
> 
> header-value        = *header-text-atom
> header-text-atom    = "\n" | "\c" | "\\" | <any OCTET except LF or ":" or "\">
> 
> 
> I should have been paying more attention during the 1.1 development process.
> 
> Regards,
>   Tony
> 
> _______________________________________________
> rabbitmq-discuss mailing list
> rabbitmq-discuss at lists.rabbitmq.com
> https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss