Re: [paho-dev] On Paho's opinionated URI validation

[Dave Locke <locke@xxxxxxxxxx>]

The current OASIS MQTT specification covers
what flows "on the wire"  and not the programming interface.
Given its remit, it is not the right place to standardise the URI.  
That said having consistency across implementations for the URI is a good
idea.   It will need to cover the different transports including  
TCP, SSL/TLS, WebSockets and WebSockets with SSL.   The approach taken
in the Paho clients has been to assume that the protocol is MQTT and then
the URI is the transport that MQTT will flow across e.g. TCP://, SSL://
 WS://, WSS://     If this was standardised is this enough
or is there a need to not assume anything and have a uri's like MQTT://
 (for TCP)  MQTTS:// (TCP with SSL/TLS)   MQTTWS://  (Websockets
...)  ? 


All the best 
Dave



From:      
 Paul Fremantle <paul.fremantle@xxxxxxxxxx>
To:      
 General development
discussions for paho project <paho-dev@xxxxxxxxxxx>
Date:      
 04/11/2013 21:21
Subject:    
   Re: [paho-dev]
On Paho's opinionated URI validation
Sent by:    
   paho-dev-bounces@xxxxxxxxxxx

---

Andy

I agree this is something that should be standardised.

Paul


On 4 November 2013 19:39, Andy Piper <andypiperuk@xxxxxxxxx>
wrote:
I think parts of those changes are sensible. You're not
missing anything, this is just code that could be improved. I think making
tcp:// and mqtt:// synonyms makes some sense.

One thing I would note though, since MQTT-S (soon to be
MQTT-SN) exists, you may want to rethink the mqtts:// as an alias for ssl://
... that's a little more tricky.

In general the Paho Java client code for URI handling
is very brittle right now. I raised this as a bug a while ago https://bugs.eclipse.org/bugs/show_bug.cgi?id=405721
and I know Al made some initial changes.

One thing I would like to ensure is that whatever scheme
is adopted, and your patch may contribute to changing (probably easiest
to attach a patch to the existing Bugzilla report)... we should try to
ensure that all of our language clients are equally capable of understanding
that scheme.

There's no standard MQTT scheme today, although IIRC once
upon a time, one of the IBM products did informally support one. I know
that some of the other third-party clients out there have adopted their
own, too (mqtt+nio etc). There is an IANA-assigned port number but currently
officially listed by an old name (1883 = ibm-mqisdp,
precursor name to MQTT, 8883 = secure-mqtt .... there's also
mqtt listed as assigned to AndySC without a port number, but with a TXT
record |Defined TXT keys: topics=<open topic to subscribe to for information>,
eg topic=/info|). In my opinion, definitely something
to be proposed for wrapping up in the OASIS work.

Thanks for raising this - useful as well in the light
of the MQTT Interop testing discussion in another thread!

Andy





On Mon, Nov 4, 2013 at 7:00 PM, Michael Klishin <michael.s.klishin@xxxxxxxxx>
wrote:
I maintain Machine Head [1] and work on a project that
provides
applications with URI's to connect to an MQTT capable broker
(RabbitMQ).

I generally find Paho Java a pretty nice library but a
couple of things
stand out as odd usability issues.

One of them is URI validation that Paho performs. Here's the exact
code that does it:

https://gist.github.com/michaelklishin/71eaeb702a7af4a20c41

This raises two questions:

 * Why is there no standard mqtt:// scheme? This
is very inconvenient for e.g. PaaS service providers that provide URI's
that developers should use.
 * Why won't Paho accept tcp://127.0.0.1:1883
but not tcp://127.0.0.1:1883/
(with a trailing slash)

tcp://127.0.0.1:1883
and tcp://127.0.0.1:1883/
are technically two different URI's but it's
pretty easy to see how a user may provide either and expect
the same result.

Am I missing something? Would a patch that makes Paho
treat
both "" and "/" as valid paths be
accepted? What about a patch
that makes "mqtt" an alias for "tcp" and "mqtts"
an alias for "ssl"?
Just to make it clear: both will not introduce any breaking changes.

Thank you.
-- 
MK

http://github.com/michaelklishin
http://twitter.com/michaelklishin

_______________________________________________
paho-dev mailing list
paho-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/paho-dev




-- 
Andy Piper | Kingston upon Thames, London
(UK)
blog: http://andypiper.co.uk
  |   skype: andypiperuk
twitter: @andypiper  |  images: http://www.flickr.com/photos/andypiper

_______________________________________________
paho-dev mailing list
paho-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/paho-dev




-- 
Paul Fremantle
Part-time PhD student - School of Computing
email: paul.fremantle@xxxxxxxxxx,
paul@xxxxxxxxxxxxx
twitter: pzfreo / skype: paulfremantle / blog: http://pzf.fremantle.org
CTO and Co-Founder, WSO2
OASIS WS-RX TC Co-chair, Apache Member
07740 199 729_______________________________________________
paho-dev mailing list
paho-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/paho-dev


Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598. 
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6
3AU