Hey Scott, As of 3.5 equinox.common should come out of the box with Virgo as p2 has dependency to it too. You’ll notice the bug doesn’t provide much content – that will change once I push the baseline of the p2 integration and build from then on for the remaining features. I have the integration prepared locally and it’s looking good, e.g. the startup time on Windows was decreased more than twice… but I’m waiting on two p2 enhancements [1] [2] to get resolved. Without them the code won’t be building and working properly. In the end if it turns out these are taking a lot of time to resolve we’ll just update EBR with the working versions until an official build is ready and start producing 3.5 alphas. Do you have the update site of the ECF's remote services? I’d be happy to try it. Best Regards Bobby From: virgo-dev-bounces@xxxxxxxxxxx [mailto:virgo-dev-bounces@xxxxxxxxxxx] On Behalf Of Hristo Iliev Sent: Friday, September 23, 2011 10:15 AM To: Virgo Project Subject: Re: [virgo-dev] Virgo and ECF remote services Hi Scott,
The bug 333174 was fixed in 3.0.0, so equinox.common should be able to run just fine.
From Virgo's point of view all needed bundles should be simply put in repository/usr to allow dependency resolving. So if ECF (or some dependent bundle) needs equinox.runtime the users will have to provide it in Virgo's repository. Then all bundles needed by ECF can be described in a plan or we can go for a library.
As for p2 there is a bug 343543 detailing the work there, but Borislav can provide more info on the topic.
Regards, Hristo Iliev On 22 September 2011 19:56, Scott Lewis <slewis@xxxxxxxxxxxxx> wrote: Greetings Virgo folks,
Since this thread was initially discussed, much has taken place with both Virgo and ECF. From ECF's side, we've had two maintenance releases of the ECF implementation of the Remote Service Admin (chap 122 from OSGi Enterprise spec)...i.e. 3.5.1/Indigo and 3.5.2 (August). We've also acquired the OSGi RSA tck via EF, and run our implementation through the tck (and made small changes/bug fixes to verify compliance). As well ECF also now has a remote services distribution that runs on Felix (and possibly other frameworks...we don't currently have the resources to do full testing on others).
When we last tried to run ECF's remote services implementation on Virgo, I recall that there was some issue with not being able to use/include org.eclipse.equinox.common (there was a bug...see below). Note that ECF remote services has compile-time dependencies on org.eclipse.equinox.common but *not* org.eclipse.core.runtime. Is this likely to still be an issue/problem?
We have seen several community requests on the ecf-dev mailing list for running ECF's OSGi remote services (and RSA) on Virgo...and would like to make this as easy as possible for community members.
A second issue: ECF's primary distribution is a p2 repo...and I think it would be very nice if an Equinox server (with or without Virgo) could be dynamically updated to add ECF remote services, Virgo, and/or both using p2 (on a server with p2 installed). Over the past few months I have not tracked closely Virgo's progress WRT p2 integration...is there a specific bug/enhancement tracking that Virgo work? Thanksinadvance for that.
Scott
On 12/29/2010 12:09 PM, Hristo Iliev wrote:
Hi,
Peter already found us [1] :) I'm not sure if he managed to deal with Virgo and equinox common - no answer from him, but if he or someone else needs assistance we are here (forums+mailing list).
Virgo does not include remoteserviceadmin in its kernel or web distributions, but I guess you can try running it on top and share the results. The topic sounds interesting.
Regards, Hristo Iliev
[1] http://dev.eclipse.org/mhonarc/lists/virgo-dev/msg00637.html
On 29 December 2010 18:55, Scott Lewis <slewis@xxxxxxxxxxxxx> wrote: Hi Hristo,
On 12/28/2010 10:21 PM, Hristo Iliev wrote:
Hi Scott,
Virgo has no problems resolving or deploying the equinox.common bundle. These operations are working as expected.
As you perhaps already saw [1] the bundle has lazy activation policy which causes confusion since its status is reported as LAZY. This status appears to be (thanks to Glyn for the explanation) a subset of STARTING. According to the specification this is the expected behavior until the bundle is activated by class loading.
The issue for which I created bug 333174 is that removal of the equinox.common bundle triggers a stop, which fails (as expected) if the bundle did not reach STARTED. There is no real problem here since the removal is ok as well.
So in short - no tricks should be needed to have equinox.common (or ECF) running with Virgo. Of course if you have some questions or issues running ECF in Virgo please come back to us - we'll be glad to help :) I guess I still don't understand (since I haven't experienced it myself), why folks like Peter [1] are having trouble with getting ECF+equinox common running on Virgo. Do you think it's likely some consequence of 333147? Should I direct Peter and other folks over to this mailing list for assistance on these issues?
BTW...I'm finishing up the ECF impl of the OSGi Remote Service Admin spec and am keen on using it on Virgo also. Does Virgo currently distribute the OSGi remote service admin classes? (i.e. org.osgi.service.remoteserviceadmin.*).
Thanks,
Scott
[1] http://dev.eclipse.org/mhonarc/lists/ecf-dev/msg04694.html _______________________________________________ virgo-dev mailing list virgo-dev@xxxxxxxxxxx https://dev.eclipse.org/mailman/listinfo/virgo-dev
_______________________________________________ virgo-dev mailing list virgo-dev@xxxxxxxxxxx https://dev.eclipse.org/mailman/listinfo/virgo-dev _______________________________________________ virgo-dev mailing list virgo-dev@xxxxxxxxxxx https://dev.eclipse.org/mailman/listinfo/virgo-dev
|