Stefan reported a problem with setting message
length in the v0.7 decoder.  Now the version
is 0.7a which fixes this problem:

 http://abelian.org/ebnaut/software_ms.shtml

Stefan wrote:

Probably that's a transmission for Paul only:

> f = 8270.100000 Hz
> Start time: 05.May.2017   05:30:00 UTC
> Symbol period: 10 s
> Characters: 100
> CRC bits: 0
> Coding 8K19A

Needs about 5 Gbyte RAM to decode when using list length
of one million, about 7 Gbyte for two million.

It's likely that using the 8K19A instead of the stronger
16K21A will work better for some given available RAM,
because the small loss of coding gain in the convolutional
code can be more than made up by the use of a longer
list.

An interesting 'feature' of longer messages is that a
false decode is often only partly false - you may find
most of the message is correct but there is a garbled
section.

--
Paul Nicholson
--