Cyrille Giquello wrote:
Hi blasters,
I try to make some Flash Applet clients for xmlBlaster.
Some problems appears.
- Could not implements subcribtions (Flash could not listen on soket).
- Could not get only new messages if no client's callback (XmlBlaster
interface get()).
I think there is one way to get a full compatible Flash / XmlBlaster
communication:
Flash can be connected to a server with a Socket like protocol. It is
a XmlSocket class.
Here is a peace of the Flash documentation say (from
http://www.macromedia.com/support/flash/action_scripts/actionscript_dictionary/actionscript_dictionary860.html)
The protocol is as follows:
- XML messages are sent over a full-duplex TCP/IP stream socket
connection.
- Each XML message is a complete XML document, terminated by a zero byte.
- An unlimited number of XML messages can be sent and received over a
single XMLSocket connection.
Perhaps it is not so hard to implements that protocol in XmlBlaster.
It looks like XmlBlaster's Socket plugin, with xml message against
binary message.
It shouldn't be too hard to use the flash XMLSocket.
In xmlBlaster (current svn) we have two ways to serialize messages and
to unpack them
again, one is the XML markup of the xmlBlaster XmlScriptParser framework
and the other
is the more binary SOCKET (XbfParser) formatting, see
xmlBlaster/src/java/org/xmlBlaster/util/xbformat/XmlScriptParser.java
(http://www.xmlblaster.org/xmlBlaster/doc/requirements/client.script.html)
xmlBlaster/src/java/org/xmlBlaster/util/xbformat/XbfParser.java
(http://www.xmlblaster.org/xmlBlaster/doc/requirements/protocol.socket.html)
We use both pluggable in our 'email' protocol layer, the XbfParser for
the SOCKET protocol,
so they are well tested.
regards
Marcel
Is that new protocol plugin could be use with other technology like
Ajax ?
Do you think this new protocol should be pertinent for XmlBlaster's
futur ?
Do you think that it could be hard to implement it ?
Has some of you can do it in one or two days ? If not, I've to do it,
but I'm not a Java agile programmer, so I will be very slow ...
thanks for discusses,
cyrille.
