[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Re: [equinox-dev] Package access across plug-ins...
Well, I feel stupid, I have attempted to reconstruct this with a trivial
example I could upload, and I can't. I'd swear it was broken, but it has
since started working. Using the commands you gave me, I managed to get my
plug-ins and fragments working together in the example I really care about.
So that was very helpful.
Now, I'm having problems, because the helpful suggestions about using
DevClassPathHelper are less useful with fragments. The DevClassPathHelper
doesn't report back the directories used by the fragment, but the Extension
Point successfully hides bundle-id of the fragment. So I can't use the
osgi.dev properties file to look it up without getting the bundle id of the
fragment from somewhere.
The fragments are inside of the AbstractBundle, but it's hidden behind a
If the fragment and the plug-in use the same directories it all works
(because overlay of bundles works great), but if the fragments use a
different directory it all stops working. I'm getting fairly close to just
using an extension point to enumerate the directories by hand. I can fix up
the auto-detection later if I find it a good way to do it.
A BundleListener might be useful, as performance isn't very important.
If you can connect a fragment to the host bundle thru public methods, it
might work out.
Just wanted to be sure and say thank you for all of your help, you guys
have been very helpful in getting my little project on track.
From: Jeff McAffer <Jeff_McAffer@xxxxxxxxxx>
Reply-To: Equinox development mailing list <equinox-dev@xxxxxxxxxxx>
To: Equinox development mailing list <equinox-dev@xxxxxxxxxxx>
Subject: Re: Re: [equinox-dev] Package access across plug-ins...
Date: Sun, 22 Oct 2006 14:35:08 -0400
Kirby wrote on 10/22/2006 02:18:59 PM:
> My current problem is finding a
> canonical way to say "Always load this fragment".
> Right now, I'm using:
> (| (osgi.ws=gtk) (!(osgi.ws=gtk)))
> Which is a tautology, it's just not as obvious as I'd like it to be.
> Leaving it blank causes it to filter out, not filter in. It works just
> on my "win32" platform. If anybody has an answer for that, it'd be
> knowing about. I'm trying to track down the code that does the parsing
> right now.
Hmmm, this feels like a bug/issue to me. To clarify, are you saying that
if you omit the Eclipse-PlatformFilter header from the fragment manifest
that the fragment is somehow disabled? If so, please run with -console
and use "ss" and "diag NN" where NN is the bundle id number of the
fragment bundle. The fragment should be RESOLVED and should show the host
bundle id as "master". If it is just INSTALLED then diag should say why
it is not RESOLVED. It may be worthwhile opening a bug report if the
fragment really is not getting resolved.
equinox-dev mailing list
Get FREE company branded e-mail accounts and business Web site from
Microsoft Office Live