|
Boost : |
From: Chris Thomasson (cristom_at_[hidden])
Date: 2006-10-24 08:15:36
"Giovanni Piero Deretta" <gpderetta_at_[hidden]> wrote in message
news:1439aad50610240210n2c9d89cdie5e462fc5f26075e_at_mail.gmail.com...
> Ok, now the tricky question,
>
> Judiging from the comments on comp.programming.threads, looks like
> almost everything that falls under lock free is patented.
Not everything... yet...
;^)
> Is the refcount code definitely patent encumbered or there is a
> reasonable chance that it could be used without falling in the patent
> swamp?
http://groups.google.com/group/comp.lang.c++.moderated/msg/744d1fcf97fa9dd9
So far so good. I have recently snooped around some patents and patent
applications and have not found anything that claims my prior art wrt the my
refcount algorithm; looks like this is going to work out just fine for
Boost.
> What about the rest of appcore (I would love to use a lockfree thread
> queue)?
Well, the current version of AppCore includes a full-blown SMR
implementation; IBM has a patent application out on it.
http://groups.google.com/group/comp.programming.threads/browse_frm/thread/dc7cea8cb012cdf5
So, I need to strip it completely out of my library, then alls that's left
are my inventions, this of course includes the queue...
> And then, what about the license? I suppose you have the intention of
> releasing at least the refcount code under the boost license,
Yup.
> what about the rest of appcore?
Humm... Not "exactly" sure yet... A commercial library of mine, called
vZOOM, makes use of a couple of my algorithms in AppCore. I currently have a
patent application out on vZOOM and it's private at this point. I only
released a particular version of it to Sun for their CoolThreads contest;
it's one of the four finalists:
https://coolthreads.dev.java.net/
So, that could be a little tricky... I really need to make up my mind
here... I will have an answer for you fairly shortly... Okay?
;^)
> other than that, great job,
Thank You! :^)
> I'm definitely not an expert in lock free
> programming and wouldn't touch the implementation not even with a 100
> meters staff, but I would definitely be an user .
That's fine. As a user, you should be able to experience the enhanced
scalability, throughput and performance benefits' that some cleverly
thought-out lock-free algorithms provide...
Indeed!
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk