java bindings, unix sockets and connection address

[Hynek Hanke]

On 4.9.2010 20:57, Rui Batista wrote:
> I'm in the process of adding unix socket communication to the java
> bindings. There is no native java support for unix sockets in java so
> I've used the libunixsocket-java package.
>    

Thank you!

> Regarding the choice of connection, I'm right now implementing the
> address string and speech_address environment variable handling to know
> what method to use. My question is, Shall I implement the old behaviour,
> i.g. SPEECH_SOCK or SPEECHD_HOST variables?

No. In the other libraries, these were kept for compatibility
reasons. If the Java bindings used SPEECHD_PORT, it might
be a good idea to continue to support it for some time (the
logic is that if SPEECHD_ADDRESS is not set, you would
respect SPEECHD_PORT).

> I don't thik java bindings
> will go for 0.7.1 mainly because I don't think they are well tested,
> spetially in the case of unix sockets.
>    

I think not. I think it's better if, after you finish the work on
them, we make a last release of speechd-java as a separate
thing. Then we integrate speechd-java into speechd tree and
it we be a part of the 0.8 release (or possibly 0.7.2).

> On other subject: What to do about the integration in the speechd tree
> of java bindings? For me it's good, I just need access to import them
> when the unix socket stuff is done. Other option is someone pull from a
> branch. You decide what is better.
>    

These are two separate things:

     1) I think you should have commit rights into the repository,
at least to the java bindings directory. I'll contact you privately.

     2) We should follow the review policy for the java bindings
too, so in this case I think it would be best if you make your
branch available somewhere rebased against master and some
of the other reviewers integrates it. If you would like, we can
create a semi-public git repository for you.

Does that sound OK?

Best regards,
Hynek Hanke