sikuli-driver team mailing list archive

[Joseph <question271271@xxxxxxxxxxxxxxxxxxxxx>]

Question #271271 on Sikuli changed:
https://answers.launchpad.net/sikuli/+question/271271

    Status: Answered => Solved

Joseph confirmed that the question is solved:
Thanks for the reply.

I do think that a native implementation has its merits, but I do also
recognize that the drawbacks in required development time and cross-
platform difficulties are significant. I don't blame you for having no
desire to do this, and  I wouldn't dare ask. I understand that Sikuli
has already required a great deal of development time and effort, and I
think it's a great project already. If I pursued this approach, I would
of course volunteer to do the work myself.

I currently do something like 3. with XML-RPC. However, this itself has
development time penalties in requiring the desired functionality to be
wrapped to be exposed to/from the Jython VM.

I have also pondered the benefits of a client/server architecture for
Sikuli, and regardless of how the API is implemented, I do think that's
a great approach to take.

Hypothetically, the API could remain implemented in Java, and depending
on the interface, the client application could use just about any
language, including CPython. Additionally, the vision work could be
offloaded to another machine,  which would free up client resources, and
make GPU acceleration more worthwhile.

I actually use the fourth approach right now, even though it's the least
efficient, time wise. I regularly run Sikuli on machines that don't
already have the JVM spun up, so initial startup takes time. After that,
everything works great.

-- 
You received this question notification because your team Sikuli Drivers
is an answer contact for Sikuli.