On 18 February 2010 00:43, David Jencks <[email protected]> wrote: > > On Feb 17, 2010, at 10:22 AM, Alasdair Nottingham wrote: > >> Hi, >> >> I've delivered a fix to the problem I saw in ARIES-174, so it now >> correctly resolves against the Application-Content rather than the >> bundles in the archive. This does reveal/introduce a different problem >> in that we don't resolve against the bundles in the archive, but I'll >> look into that tomorrow and send an update. > > Is the behavior you expect (resolving against the application-content) > specified somewhere?
Yes, it is specified here: http://incubator.apache.org/aries/applications.html > > What would happen if we stuff resources for all of the application content > (like you do) and the bundles in the archive (like I did) into the resolver? > I was under the (untested) impression that obr would resolve the supplied > resources against each other. > It attempts to resolve all the resources that have been added to the resolver. So Let's assume we have the following: archive: Application-Content: a, b, z a.jar b.jar c.jar d.jar e.jar a depends on b b depends on c d depends on q e depends on r z depends on w If I add a, b, z in as resources I would end up with a, b, c, z, w. If I add the bundles in the Application-Content, and the bundles inside the archive I would end up with: a, b, c, d, e, q, r, w, z which would result in too much being provisioned, and content that wasn't asked for. Note if no Application-Content is specified it would default to a, b, c, d, e I'm working on a fix today that will correctly resolve against the bundles in the archive. >> >> In addition I managed to get the failing itest running, so I've >> committed a fix that should fix that. >> >> We really need to get some debug into the resolver, I'm not familiar >> with SLF4J though, perhaps we should have a page on the wiki to >> describe how we should use it? >> > > I tend to prefer jpda debugging to logging except to explain errors in > detail.... but what kind of logging are you thinking of? For me slf4j type > logging is so ingrained I don't know how to explain how to use it. > I agree I prefer jpda, but it often isn't an option. I frequently just get a set of trace/log files from customers using the code and I need to debug from that as the customer's application/setup is not available. > I'll put a logger in and a dumb example use, maybe that will help. > Thanks > thanks > david jencks > >> Alasdair >> >> On 17 February 2010 10:56, Alasdair Nottingham <[email protected]> wrote: >>> >>> I've found a bug in the resolver. The resolver doesn't resolve the >>> application content, but it resolves the bundles contained inside the >>> application archive. I'm working on a fix and I'll commit it under >>> ARIES-174. >>> >>> Alasdair >>> >>> On 17 February 2010 10:11, Alasdair Nottingham <[email protected]> wrote: >>>> >>>> I would advise against exporting any implementation package from a >>>> bundle. >>>> >>>> Alasdair >>>> >>>> On 16 February 2010 20:00, David Jencks <[email protected]> wrote: >>>>> >>>>> I committed a basic OBR-backed AriesApplicationResolver, see ARIES-174. >>>>> >>>>> To get the itest to run it looks like you need to build >>>>> bundlerepository >>>>> locally, I'm not quite sure why. >>>>> >>>>> I'm seeing the new MinimumImportsTest itest fail, I haven't looked into >>>>> why. >>>>> >>>>> This might be an overly naive implementation, so if I missed big pieces >>>>> of >>>>> functionality and someone wants to start over I won't mind :-). >>>>> >>>>> I'd also appreciate some advice from more-osgi-experienced developers >>>>> as to >>>>> whether it's appropriate to export the main resolver class or if it >>>>> should >>>>> be hidden in the impl directory. >>>>> >>>>> thanks >>>>> david jencks >>>>> >>>> >>>> >>>> >>>> -- >>>> Alasdair Nottingham >>>> [email protected] >>>> >>> >>> >>> >>> -- >>> Alasdair Nottingham >>> [email protected] >>> >> >> >> >> -- >> Alasdair Nottingham >> [email protected] > > -- Alasdair Nottingham [email protected]
