Matt Fletcher wrote:
I've discussed this a bit with headius and others in IRC over the last
two days. It sounds like there may be a better way out: JFFI. I do not
know much about the JFFI project since I cannot find any information
about it on the web. But headius is saying that JFFI is not dependent on
JNA and it could easily substitute for where JNA is being used right now.
So I have a couple of questions:
1) Is JFFI going to be as or less restrictive than CPL for commercial
applications?
2) Can JFFI truly replace JNA JRuby?
3) What can I do to help with the conversion?
4) If JFFI can't be used or isn't ready yet, am I in for a world of pain
when I go to produce a JNA-less JRuby?
These are largely questions for Wayne, since JFFI is his baby. If he'd
willing to add a more BSD/MIT-like license it might solve your long term
issues.
In the short term, we ought to look at a way to decouple JNA from JRuby
proper. Much of that has already been done through the jna-posix
project, which would only take a little work to rewire to fall back on
the Java-based impl when JNA isn't present. But there are also other
places we use JNA inside JRuby that should probably be made to fall back
or disable themselves when JNA isn't present. UNIX sockets come to mind,
for example.
- Charlie
---------------------------------------------------------------------
To unsubscribe from this list, please visit:
http://xircles.codehaus.org/manage_email
