I believe a bug was fixed for this. Can you try with 2.3.0.Beta1? I'm looking for the related JIRA in the mean-time.
works better , thanks !
I believe I have seen a case about someone asking us to set the Text there instead of the encoding. (In case of TextMessages). however.. there are two issues:
I - On core level everything is byte-array (on the server's side).
II - We could infer the message-type property that is on the message, (by decoding it first), but then there's an issue with CharSets... How to preserve it? This one here will work fine whatever the chartset. (it's just plain bytes.. and they get preserved).
You're right - my first suspicion was also an encoding-issue, and I lost some time with this f...cking I/O-API to make ensure everything is UTF-8 ...
To working only with bytes is enough at this level.
An other question - is it possible to reuse the message id ? I see your UUID has no "parsing" function (yet). But so we can make sure, that messages are imported only once.
that was done on purpose... the messageID should be assigned at the time of the import. The importing will be a basic producer, so the messageIDs will be re-generated.
When we came up with this, we even though about inter operability with brokers.