| Hi,
FWIW, I agree with David. The qualifier really ought to signal a functional change in the bundle. That might not necessarily be just in the executable code, of course: changes to the values of localized strings or OSGi metadata would warrant new qualifiers, as would signing changes if they had a manifest performance impact.
cW
On 3-Feb-09, at 12:01 AM, David M Williams wrote: > Discussion topic: In either case, it seems the binary content of
> the Orbit bundles has changed but the qualifiers have remained the
> same. Should all the orbit cvs bundles be tagged with new versions
> to change the qualifiers to notify consuming clients that there is
> new content? Otherwise, there isn't a way to signify to the
> community that the content of the jars in the two builds are not
> identical. I would propose that this should be done for the next
> milestone, not M5.
Offhand, I think there'd be no need to do this, since the _executable_ code is not changed.
For example, in the case of install, if someone had the old one installed,
there really is no need for them to download and install a new one, just because the new one is signed
with a different certificate.
We went through a similar discussion long ago when the issue was "signed" vs. "not signed", and
concluded there was no reason to change the qualifier even in that case.
I am, of course, open to discussion and counter arguments.
orbit-dev mailing list
orbit-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/orbit-dev
-- Christian W. Damus Senior Software Developer, Zeligsoft Inc. Component Lead, Eclipse MDT OCL and EMF-QTV
|
