Status of Boincoid

[Bill]

Responding to Sam's message... I haven't looked @ Boincoid in a
while. I thought I pushed my code modifications back up to
SourceForge when I was initially working on this, but perhaps I
forgot.

Something to note here is that Boincoid was not a fully functional
BOINC client, though it was close --- at least when I worked with it
in 2009-2010. Although it demonstrated the ability to process SETI
work units on Android, it did so only with a sample SETI work unit.
It did not fully implement the network services required to
communicate with the BOINC SETI server. Doing this would have
required coordination and cooperation with the BOINC SETI project
team. Perhaps someone has taken the Boincoid project further since
then, but if so, judging from the source code dates the project has
not been updated.

After getting Boincoid running on the Android Simulator in Eclipse,
and realizing the limitations inherent to the approach of converting
BOINC (and then each individual project) to Java, I switched gears and
wrote an x86 virtual machine in Python that I used to simulate a BOINC
client and run SETI on an iPhone 3. Here's a link to an abstract of
my MS Thesis, which was done on this subject: http://gradworks.umi.com/14/70/1470932.html
Using the x86 virtual machine, I was able to implement network
services that allowed me to successfully download work units and
upload results to the SETI@Home server, however the x86 virtual
machine was not a complete implementation, so not all instructions
required to run the full BOINC client were available. I can share
more detail if it will help, or is of interest.

Regarding Boincoid, I'll take a look at what code I've got, and
whether it will still compile and run. I've since loaded many new
versions of Android and the Android simulator, so I'm not sure whether
I can get it to compile and run again or not.

-Bill

[Oded Ben-Dov]

Bill,

thanks for your involvement in the project and the great summary below.

Boincoid is still not complete, with gaps as Bill mentioned and much debugging and packaging still to do.

Today one could also wrap the engine in a java wrapper and keep the existing code. Haven't tried it, but it might a be a quicker fix.

I keep believing that there's an advantage to a pure java engine, that works with non-pure java applications. But maybe it's just clinging nostalgically to my code :)

Cheers,

Oded.