jeff at licquia.org
Tue Apr 22 07:21:25 PDT 2008
Joseph Kowalski wrote:
> I think the previous threads and our conference call seemed to close the
> issue that Java is part is intended to be part of the LSB.
We talked about this at the LF Summit earlier this month. Observations
from there follow.
> 1 Java interfaces are to be exported by the LSB (as much as a
> specification can actually export anything). For the most part, this
> will be "specification by reference", but a number of details will be
> LSB specific, such as installation locations and supported CLI utilities.
> 2 A JVM contained in a LSB compliant distro does not necessarily
> need to be otherwise LSB compliant, meaning the the distro JVM can
> depend upon non-LSB interfaces. In fact, this is a feature because it
> allows distros greater freedom to innovate, probably along the
> performance axis. (Possibly delta installation locations, but that's a
> lower level topic.)
> 3 Of course, a non-bundled JVM (as in java.sun) may or may not be
> otherwise LSB compliant. However, non-bundled JVMs which can claim LSB
> conformance should have a marketing advantage.
> 3a It is likely that most distros will desire to use JVMs which can
> assert they are LSB compliant. At this point in time, most distros
> would want the "most easily supported" (and robust) JVM they can get.
I'm not sure this is correct. Distro-provided JVMs may have good reason
to break LSB application compatibility, as you note above.
Of course, we want to eliminate gratuitous incompatibilities. It would
also be nice to enable conditional compilation, so you can build a JVM
in "LSB mode", with a slow but standards-compliant implementation, or
"native mode", with the volume turned up to 11. That would help with
> 4 There is nothing special about applications which embedded a JVM
> (or carry along a private JVM --- yecch, but not too rare) with respect
> to the LSB. It would probably be prohibitally difficult to embed a
> non-LSB compliant JVM (as in #3 above) into an application wishing to
> assert LSB compliance.
> If we can agree on these 4 bullets (massaged as appropriate) we can
> proceed to lower levels of specification. If not, well, let's talk (but
> I think we had rough agreement on the con-call).
I think the plan is to assume that we can rely on Sun for most of the
Java specs, testing, etc. via the published specs and a to-be-announced
certification program for Linux distributions.
As a fallback, in case Sun runs into more issues that delay the cert
work, the Java specification can be released as a trial-use module.
More information about the lsb-discuss