[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [p2-dev] installing bundle fragments
- From: Igor Fedorenko <ifedorenko@xxxxxxxxxxxx>
- Date: Wed, 08 Apr 2009 11:56:45 -0400
- Delivered-to: firstname.lastname@example.org
- User-agent: Thunderbird 18.104.22.168 (X11/20090318)
Pascal Rapicault wrote:
This is the expected behaviour in 3.4 and 3.5.
Does this mean you expect this to change in 3.6?
The problem with having the IU for the host bundle expressing a
requirement on an IU providing a " "fragment of host" capability" is
that at install time this can result in additional things to be
installed and that are not necessarily desired since you would pull in
That said, in the particular case of SWT, given that the host is useless
since it is empty (no API there and all the code being in the fragment)
having the IU for the host express a requirement on a
"org.eclipse.swt.fragment" capability and have every fragment provide it
could do the work. Feel free to open a bug against SWT to that extent.
SWT bundle/fragments apart, what is the recommended way of installing
desired fragments with P2? Bundle fragments are a common place but the
only way to install them appears to be explicitly listing them in group
IUs. How do you include ECF SSL fragments in P2 application, for
example? Or maybe you can suggest another example I can look at.
Inactive hide details for Igor Fedorenko ---04/07/2009 03:43:16
PM---Hello, I was under impression that P2 is expected to find Igor
Fedorenko ---04/07/2009 03:43:16 PM---Hello, I was under impression that
P2 is expected to find and install any
Igor Fedorenko <ifedorenko@xxxxxxxxxxxx>
P2 developer discussions <p2-dev@xxxxxxxxxxx>
04/07/2009 03:43 PM
[p2-dev] installing bundle fragments
I was under impression that P2 is expected to find and install any
bundle fragments applicable to the target runtime environment. So if,
for example, I tell it to install org.eclipse.swt on linux/gtk/x86_64 it
should install applicable swt fragment too. This does not seem to be the
case, at least not with director app from 3.5M6. I see that P2 is able
to filter out inapplicable fragments that are referenced through
features, but does not actively try to "find" additional fragments. Is
this expected behaviour?