Hi Paul, It appears that if a Fragment has an osgi.serviceloader Capability, the intention is that the host bundle must have a class that can be service loaded. The commit at https://github.com/apache/aries/commit/76f39f2#diff-aed8297fc5209027131b1f54544e00eb7614070832051a29561931ed6b933ab7 seems relevant. Furthermore, examples of fragments in the Aries repository appear to confirm this theory. This would imply that, without shading Groovy into a fat JAR, it won't be possible to get this working. Fat Jar is working, just tested. Or, another option, moving all services to a separate bundle. However, I suspect this also won't be feasible for every service. ...Any suggestions?
br. Matthias On Tue, 25 Feb 2025 at 22:52, Matthias Leinweber <m.leinwe...@datatactics.de> wrote: > Hello Paul, > > I've conducted further tests. While the FastStringFactor sometimes works > after multiple application restarts, it's mostly failing. I suspect a > problem with the mediator components. The service loader specification ( > https://docs.osgi.org/specification/osgi.cmpn/8.0.0/service.loader.html), > implemented by Aries spifly, should, as I understand it, create an OSGi > service. However, this service doesn't seem to be getting created. > Interestingly, this same missing service issue also occurs with Groovy > 3.0.x bundles. I'm quite confused by this. > > But these services are created for version 4.0.25 (without a fragment). I > assume that SPI Fly only works for started and not for resolved bundles. > However, this doesn't explain why 3.0.25 is working, nor why it sometimes > works after multiple restarts. > > I will investigate further tomorrow. > > On Tue, 25 Feb 2025 at 21:24, Matthias Leinweber < > m.leinwe...@datatactics.de> wrote: > >> Hi Paul, >> >> I've encountered some issues with the FastStringService. I'm getting an >> "Unable to load FastStringService" error and I'm trying to determine the >> cause. I vaguely recall having trouble with this class in the past, but I >> can't remember the specifics. >> >> Given these problems, I don't think this is releasable yet. >> >> Best regards, >> Matthias >> >> On Mon, 24 Feb 2025 at 20:19, Matthias Leinweber < >> m.leinwe...@datatactics.de> wrote: >> >>> Did some basic tests. Extensions seem to work and FastString Service too. >>> >>> br; >>> Matthias >>> >>> On Fri, 21 Feb 2025 at 23:15, Paul King <pa...@asert.com.au> wrote: >>> >>>> There should be 4.0.26-SNAPSHOT and 5.0.0-SNAPSHOT versions in the >>>> snapshot repo: >>>> >>>> >>>> https://repository.apache.org/content/repositories/snapshots/org/apache/groovy >>>> >>>> On Fri, Feb 21, 2025 at 10:30 PM Matthias Leinweber >>>> <m.leinwe...@datatactics.de> wrote: >>>> > >>>> > Hi Paul, >>>> > >>>> > could you assist with the maven repo containing the snapshots or do i >>>> have to build myself? >>>> > >>>> > br: >>>> > Mathtias >>>> > >>>> > >>>> > On Thu, 20 Feb 2025 at 11:40, Paul King <pa...@asert.com.au> wrote: >>>> >> >>>> >> I have pushed a fix. Once CI has finished building snapshots, if you >>>> >> can check against those, that would be great. >>>> >> >>>> >> Cheers, Paul. >>>> >> P.S. Feel free to comment further in that issue instead of here if >>>> you want. >>>> >> >>>> >> On Thu, Feb 20, 2025 at 8:32 PM Paul King <pa...@asert.com.au> >>>> wrote: >>>> >> > >>>> >> > I created this issue to track changing this for Groovy 4/5: >>>> >> > >>>> >> > https://issues.apache.org/jira/browse/GROOVY-11570 >>>> >> > >>>> >> > On Wed, Feb 12, 2025 at 11:14 PM Paul King <pa...@asert.com.au> >>>> wrote: >>>> >> > > >>>> >> > > I think we should add the FragmentHost stuff back in and get a >>>> release >>>> >> > > out in the next few weeks. >>>> >> > > >>>> >> > > On Tue, Feb 11, 2025 at 2:05 AM Matthias Leinweber >>>> >> > > <m.leinwe...@datatactics.de> wrote: >>>> >> > > > >>>> >> > > > I am not sure if someone ever tested this... Documentation >>>> says to get groovy extensions working is that the extension must be in the >>>> same class path (class loader). >>>> >> > > > >>>> >> > > > Maybe I am wrong but so fragment host (terrible name imho) is >>>> the only solution. Or shading a new jar could do the same trick. >>>> >> > > > >>>> >> > > > What do you think? >>>> >> > > > >>>> >> > > > Br >>>> >> > > > >>>> >> > > > Paul King <pa...@asert.com.au> schrieb am Mo., 10. Feb. 2025, >>>> 14:25: >>>> >> > > >> >>>> >> > > >> In the following issue we were not certain of the need for >>>> >> > > >> Fragment-Host in Groovy 4: >>>> >> > > >> >>>> >> > > >> https://issues.apache.org/jira/browse/GROOVY-9402 >>>> >> > > >> >>>> >> > > >> It could well be that it is needed. >>>> >> > > >> >>>> >> > > >> Cheers, Paul. >>>> >> > > >> >>>> >> > > >> On Mon, Feb 10, 2025 at 10:00 PM Matthias Leinweber >>>> >> > > >> <m.leinwe...@datatactics.de> wrote: >>>> >> > > >> > >>>> >> > > >> > Hi together, >>>> >> > > >> > >>>> >> > > >> > I am using groovy alot in an osgi container with camel >>>> (currently karaf.4.7.7 camel 4.9.1 and groovy 4.0.25) >>>> >> > > >> > >>>> >> > > >> > Somehow the dateutils overload for the date class seems not >>>> to work anymore. In groovy 3.0.9 the bundle was a fragment host and >>>> working.... Do I need to do something extra with groovy 4? >>>> >> > > >> > >>>> >> > > >> > Br Matthias >>>> > >>>> > >>>> > >>>> > >>>> >>> >>> >>> >> >> > >
