Ok, it took me some time to actually implement the Scala based OSGi console.
For an introduction see:
http://incubator.apache.org/clerezza/getting-started/shell/

To try it out you may either download a recent launcher for clerezza from

https://repository.apache.org/content/repositories/snapshots/org/apache/clerezza/org.apache.clerezza.platform.launcher.tdb/0.5-incubating-SNAPSHOT/

alternatively you may install the shell manually in your container, by
installing and starting the following bundles (mvn uris to be resolved from
the repo above)

mvn:org.apache.clerezza.scala/scala-compiler-osgi/0.1-incubating-SNAPSHOT
mvn:org.apache.clerezza.scala/scala-library-osgi/0.1-incubating-SNAPSHOT
mvn:org.apache.clerezza.scala/script-engine/0.1-incubating-SNAPSHOT
mvn:org.apache.clerezza/org.apache.clerezza.shell/0.1-incubating-SNAPSHOT
mvn:org.apache.clerezza/org.apache.clerezza.shell.felixshellcommand/0.1-incubating-SNAPSHOT
(optional, allows to execute classic felix shell commands)

to allow access via ssh (default port 8022) the following bundles are
required additionally:

mvn:org.apache.clerezza/org.apache.clerezza.sshshell/0.1-incubating-SNAPSHOT
mvn:org.apache.mina/mina-core/2.0.2
mvn:org.apache.sshd/sshd-core/0.5.0
mvn:org.apache.sshd/sshd-pam/0.5.0

Cheers,
Reto


PS: I put together the lists of bundle uris above from the scala shell using
the following commands on the scala based shell:

zz>for (b <- bundleContext.getBundles; if
b.getLocation.contains("clerezza.scala") ||
b.getLocation.contains("clerezza.shell")) out println b.getLocation
zz>for (b <- bundleContext.getBundles; if b.getLocation.contains("ssh") ||
b.getLocation.contains("mina")) out println b.getLocation





On Mon, Jul 12, 2010 at 1:38 PM, Reto Bachmann-Gmuer <
[email protected]> wrote:

> I think once we have the scala console working registering a gogo command
> to access it should be quite straight forward. Still I think the independent
> scala console should best be implemented first.
>
> Cheers,
> reto
>
>
> On Fri, Jul 9, 2010 at 5:56 PM, Derek Baum <[email protected]> wrote:
>
>> I've recently been experimenting with adding groovy support for gogo, and
>> I
>> assume that scala would be similar.
>>
>> My use-case is that although gogo is scriptable, it was never intended to
>> be
>> used instead of groovy or scala.
>>
>> I therefore wanted the ability to run a full blown groovy script from
>> gogo,
>> giving the script access to my gogo session variables and an OSGi context.
>> Here's a brief look of what it can do so far:
>>
>> [Note: I'm using posh here, which is part of our commercial Nimble
>> product,
>> but this could easily be ported to gogo]
>>
>> % cat hello.groovy
>> #!/usr/bin/env posh -c groovy
>>
>> println "hello, groovy!"
>>
>> // access bundle context
>> println context.bundle
>>
>> // run gogo command
>> def b0 = session.execute("bundle 0")
>> println "b0=" + b0
>>
>> // concatenate arguments from gogo
>> def result = ""
>>
>> for (arg in this.args) {
>>  println "Argument:" + arg.class + ": " + arg;
>>  result += arg
>> }
>>
>> return result
>> %
>>
>> # now run hello.groovy script from posh/gogo
>> # note: that arguments and return value are handled correctly
>>
>> % groovy hello.groovy $HOME $SHLVL
>> hello, groovy!
>> com.paremus.posh.runtime_1.0.21.SNAPSHOT [1]
>> b0=org.eclipse.osgi_3.6.0.v20100517 [0]
>> Argument:class java.net.URI: file:/Users/derek/
>> Argument:class java.lang.Integer: 1
>> %
>> % echo $_
>> file:/Users/derek/1
>> %
>>
>>
>>
>> # now run groovy interactively from posh/gogo
>> # this is less advanced, as there is no command-line editting or
>> completion.
>> # rather than posh/gogo try to provide this, it should be provided by the
>> scripting environment
>> # so that gogo/scala shell handles completion/editting in exactly the same
>> way as the non-OSGi scala shell.
>>
>> % groovy
>> groovy$ println context.bundle(0)
>> groovy: groovy.lang.MissingMethodException: No signature of method:
>> org.eclipse.osgi.framework.internal.core.BundleContextImpl.bundle() is
>> applicable for argument types: (java.lang.Integer) values: [0]
>> Possible solutions: getBundle(), getBundle(long), getBundles(),
>> findAll(groovy.lang.Closure), find(groovy.lang.Closure),
>> use([Ljava.lang.Object;)
>> groovy$
>> groovy$ println context.getBundle(0)
>> org.eclipse.osgi_3.6.0.v20100517 [0]
>> groovy$ %
>>
>>
>> Does this have any synergies with what you're want to do, assuming we can
>> create similar functionality using scala?
>>
>> Regards,
>>
>> Derek
>>
>>
>>
>> On 7 July 2010 15:39, Richard S. Hall <[email protected]> wrote:
>>
>> > On 7/7/10 3:39, Reto Bachmann-Gmuer wrote:
>> >
>> >> On Tue, Jul 6, 2010 at 11:09 PM, Richard S. Hall<[email protected]
>> >> >wrote:
>> >>
>> >> I'm not too familiar with Scala, so pardon my ignorance.
>> >>
>> >>
>> >>> So is the proposal to have some sort of Scala-based console/shell?
>> Does
>> >>> this mean you can do Scala-based scripting and syntax? Is this
>> something
>> >>> that could simply be another shell front end for the Gogo runtime or
>> is
>> >>> it
>> >>> somehow completely different?
>> >>>
>> >>>
>> >>>
>> >> I must  admit I'm not familiar with OSGi RFC-147 so I'm not sure if it
>> >> could
>> >> be implemented on it.
>> >>
>> >> To be as attractive for people currently not using OSGi it should feel
>> as
>> >> much as possible like the scala console, this includes:
>> >> - autocompletion with tab
>> >> - multi-line expressions (afte an incomplete expression such as one
>> >> opening,
>> >> but not closing a bracket a continuation-prompt appears)
>> >>
>> >> Once we have this we can add a DSL to more easily do OSGi related tasks
>> >> (listing services and bundles, accessing services, etc. )
>> >>
>> >>
>> >
>> > Well, I still can't say I totally understand what is being proposed, but
>> > I'm not against having people work on it at Felix if other people think
>> its
>> > worthwhile. If it does ultimately blossom into a full-blown shell for
>> OSGi,
>> > then there will certainly be some overlap with Gogo, but we can always
>> look
>> > for ways to bridge the two...
>> >
>> > -> richard
>> >
>> >  Reto
>> >>
>> >>
>> >>
>> >
>>
>
>

Reply via email to