On 5/7/11 9:25, BJ Hargrave wrote:
Started means Bundle.start was called. So the bundle may be STARTING
or ACTIVE. If it is STOPPING, then Bundle.stop has been called and it
will soon be RESOLVED.
If, during refresh, a started bundle is stopped but does not
transition to RESOLVED (this could be due to an error in the activator
because it hangs), this is a corner case. A case could be made that we
fail the refresh. But I would say the bundle's activator is in error
and that the refresh proceeds. This is like stopping a bundle. When
Bundle.stop is called, errors in the activator do not prevent the
framework from moving the bundle to the RESOLVED state.
I was thinking about the case where you actually have a bundle in the
STARTING state because the activator is currently executing, in which
case I think it is reasonable that the refresh operation could fail. I
wasn't thinking about the case of lazy activation, which should always
be stoppable, so it should never cause an issue.
So, in short, I agree with BJ, my original comment was directed at the
corner case...I rarely think about lazy activation...
-> richard
--
*BJ Hargrave*
Senior Technical Staff Member, IBM
OSGi Fellow and CTO of the _OSGi Alliance_ <http://www.osgi.org/>_
[email protected]_ <mailto:[email protected]>
office: +1 386 848 1781
mobile: +1 386 848 3788
From: Ikuo Yamasaki <[email protected]>
To: OSGi Developer Mail List <[email protected]>
Date: 2011/05/07 08:55
Subject: Re: [osgi-dev] Re: Q on PackageAdmin#refreshPackages()
Sent by: [email protected]
------------------------------------------------------------------------
Hi BJ,
On Fri, 6 May 2011 22:31:46 -0400
BJ Hargrave <[email protected]> wrote:
BJ> Yes. Since PackageAdmin has been deprecated and replace by the Bundle
BJ> Wiring API, we will not be updating PackageAdmin any more.
Got it.
BJ> You list 2 suggested changes below. The first one should be to
replace
BJ> ACTIVE with the more general term "started" which can include lazy
BJ> activated bundles in the STARTING state.
I'm confused a little bit .
For me,
a. "started" means "in the state of ACTIVE".
b. a lazy activated bundle can be in the STARTING state.
Therefore the following description covers all.
>> 2 Each bundle in the graph that is in the state of STARTING, ACTIVE
or STOPPING will be stopped as described in the Bundle.stop method.
(I had miswritten "STARTING, ACTIVE or" instead of "STARTING, ACTIVE or
STOPPING")
BJ> The second one is incorrect: the
BJ> remaining steps will not be ignored. That is why it is not a
numbered list
BJ> item.
That is different from what Richard wrote:
Richard> Either the refresh operation will end up waiting for B to
become active
Richard> before stopping it and refreshing it or the entire refresh
operation
Richard> will fail because B couldn't be stopped.
Which is expected ? or it might depend on the implementation because the
spec is ambiguous.
Best regards,
Ikuo
=======
Ikuo YAMASAKI
_______________________________________________
OSGi Developer Mail List
[email protected]
https://mail.osgi.org/mailman/listinfo/osgi-dev
_______________________________________________
OSGi Developer Mail List
[email protected]
https://mail.osgi.org/mailman/listinfo/osgi-dev
_______________________________________________
OSGi Developer Mail List
[email protected]
https://mail.osgi.org/mailman/listinfo/osgi-dev
