Hi Simon,

could you please post your package on the universe? The how-to is at
http://wiki.squeak.org/squeak/5899.

Thank you

2007/10/4, Simon Holland <s.holland at open.ac.uk>:
>
> Hi All
>
> I have just uploaded a new version of OSC (Open Sound Control) which
> adds a number of enhancements and fixes. It tests all green, but please let
> us know if you find any problems.
>
> Open Sound Control is
> • used in music and sound applications  (e.g. SuperCollider),
> • used in tangible user interaction using video tracking (e.g. Tuio for
> Squeak),
> • a fast, flexible, light, easy to use, general purpose connector in cases
> where you would otherwise need to write a plugin.
>
>
> Changes in this version
> OSC has been enhanced to deal with bundles, intensive input, and the Tuio
> protocol. Various bug fixes and new unit tests have been added. No changes
> are required for clients. New features are as follows.
>
> 1 Bundles processed fully
> When several incoming messages arrive in a single bundle, or more than one
> bundle arrives at once, all messages are now parsed. This involved changes
> to the OSCParser method parse: and to the OSCServer message receive: This
> necessitated changes to three parser unit tests, and one new unit test.
>
> 2 Word boundaries treated correctly in string arguments.
> The method parseString in OSCParser now only consumes trailing nulls (after
> the null that terminates the string) if this is needed to reach a 32-bit
> word boundary, as per the OSC specification. This avoids problems where
> subsequent integers, floats etc in messages could be corrupted. A second new
> unit test was added.
>
> 3 Incoming OSC buffers resized to cope with high volume input. Incoming OSC
> packets of 940 bytes and larger are routinely sent by Reactivision. The
> incoming String and RWBinaryOrTextStream buffers in OSCServer listenOnPort:
> have been resized to cope with high volume incoming traffic.
>
> 4 OSCServer processing balanced.
> In some circumstances, the looping background UDP read thread in OSCServer
> listenOnPort: can monopolise the processor, preventing other threads from
> processing the input in timely fashion. The method receive: in OSCServer has
> been given a delay to allow upstream processes to get adequate processor
> time. An instance variable responsePeriod measured in milliseconds has been
> added to OSCServer, and initialised in initialize to allow this delay to be
> tuned conveniently for different applications. In the case of applications
> of Reactivision, a 10 millisecond delay is a suggested starting point. This
> delay can be set to zero where not needed.
>
> 5 Bug fix to prevent exception in parseHeader in OSCParser
> OSCParser was erroneously testing for a character in a binary stream and
> throwing an exception. The method parseHeader in OSCParser has been amended
> to put the stream into ascii mode at the crucial point.
>
> 6 Minimal measure to ignore blobs more gracefully.
> A dummy method has been added to ignore blobs, rather than throwing an
> exception. If a blob message is spuriously recognised as a result of a
> glitch (which occasionally happens), it is very likely to be the last or
> only message in the bundle, so that this rudimentary measure will allow it
> to be discarded without causing an exception. A future version could be
> coded to handle blobs properly, or raise and handle an appropriate
> custom-designed exception.
>
> Summary of changes
> a) One new, and two revised, instance methods for OSCParser
> parseHeader
> parseString
> parseBlob
>
> b) One revised class method for the OSCParser
> parse:
>
> c) Three revised instance methods for OSCServer.
> initialize
> listenOnPort:
> receive:
>
> d) Five revised unit tests
> OSCParserTest testParseIncreaseVolumeMessage
> OSCParserTest testParseExampleBerkely3Message
> OSCParserTest testParse
> OSCMessageTest testByteArrayDecreaseVolume
> OSCServerTest testSendAndReceive
>
> The first three tests are required by the representation change introduced
> by (1) above. In the fourth test, the literal array used to represent 0.2
> using an IEEE32Bit Float representation has been corrected. In the fifth
> test (which fails only intermittently as it involves a timing delay), the
> wait period before reading has been changed from 1 millisecond to 30
> milliseconds.
>
> e) Two new unit tests OSCParserTest2 and OSCParserTest3 test the integrity
> of enhancements/fixes 1-3 and 5. These tests would benefit from being split
> into several smaller tests, but they are adequate to identify the presence
> or absence of the relevant behaviours.
>
> Simon Holland
>
>
> --
> View this message in context: http://www.nabble.com/-ANN--OSC-%28Open-Sound-Control-tf4568812.html#a13039836
> Sent from the Squeak - Dev mailing list archive at Nabble.com.
>
>
>


-- 
Damien Cassou