[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ecf-dev] How to consume HelloWorld remote OSGI Samples in eclipse platform
- From: Pradeep Fernando <pradeepfn@xxxxxxxxx>
- Date: Sat, 23 Apr 2011 12:28:16 +0530
- Delivered-to: email@example.com
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type:content-transfer-encoding; b=ngDSbTj9LEDBqsUyob+Bw37hV5VFEqw46bjHUydYiW0HZZALnKzswcdtMwDXDmh0fj ogIWEwaSfYtaLe3xEevSFFCOSV/ETQdsvNbFViZaAubZKZoSDTcrLL1T8PBOu5R/IwYv ulnFEDsJSxaxE+n9ROsVHT99tu++kM6Ivzuis=
On Fri, Apr 22, 2011 at 11:02 PM, Scott Lewis <slewis@xxxxxxxxxxxxx> wrote:
> Hi Harshana,
> On 4/21/2011 9:36 PM, Harshana Eranga Martin wrote:
> <stuff deleted>
> Pradeep has tried this on Equinox.
>> ÂI've never seen it on Equinox. ÂShould it be considered a framework bug?
>> Is there something we can do...with the examples or RSA...to prevent this?
> I think we should mention this in the user guide for samples. For example,
> if we are trying a sample, what set of bundles to be started and
> ÂtheirÂprerequisites. That will certainly help new comers to ECF IMHO.
> I agree about adding this to the 'Running the Example Hello Service Host'
> section of the appropriate wiki page .Â Would you (Harshana) be able to
> add this info for one combination of discovery and distribution providers?
> Note that the product configurations that are referenced in that section
> already have the necessary bundle autostartÂ and startlevel information in
> them (in the configuration section of the product config)...and this could
> be copied from a given product config and put into the example docs.
> But separate from the issue of more documentation on running/starting the
> examples......I don't understand why it's necessary...under any
> conditions...to explicitly start the
> org.eclipse.ecf.examples.remoteservices.hello bundle.Â It seems to me that
> it should be started upon attempt to load any of its interface classes (e.g.
> IHello)...and in fact this is what happens when I run/start the hello
> examples on Equinox.ÂÂ I would like to try to figure out why it's necessary
> to start this bundle explicitly under some conditions...and if possible fix
> any underlying issues (in ECF or Equinox or Felix or ...)
> So...Pradeep or Harshana and/or others...would you please describe the
> conditions (in terms of the framework, the way the framework is started,
> what bundle state things are in, etc), that lead to being able to reproduce
> the bundle state exception that Pradeep reported?Â It would be much
> appreciated...and hopefully by doing this we can completely avoid this issue
> in the future.
In the case of hello world example that make use of activators/service
trackers instead of declarative services, I got that bundle state
exception issue., when i did "start
According to Markus's response then I tried to start the interface
bundle "hello.bundle". But it didnt work. Then with the help from this
forum discussion,  I figured out that we should get other bundles
in to "Active state" prior starting the
In the second run, I made sure other bundles were in Active state
before starting the
org.eclipse.ecf.remoteservices.hello.consumer.edef bundle. bundle.
However since OSGI runtime remembers the last state of bundles, in the
second run, rest of bundles including interface bundle were in Active
state AFA I can remember.
Framework : Equinox
Target environment : Helios + ECF remote services target components.
run using given .product files.
Please let me know if i can give even a small help in this regard.
Happy to contribute.
> (I would suggest putting in a new page with a link on the above page...to
> give detailed information about bundle autostart/start levels, use of Felix,
> Thanks and Regards,
>> ecf-dev mailing list
> Is this something specific to Felix?