Yes!
Maybe there is something reusable from the xstream eliminating work.
Stefan
Jeff McAffer wrote:
So far the approach has been that
ArtifactRepositories
and/or processing steps are responsible for defining how descriptors
are
written. So, in the delta case, the need to serialize an ArtifactKey
is driven by the implementation of the delta processing step. You
should therefor store the needed information any way you want. Of
course, if it turns out that alot of people need to serialize some
object
then it would be convenient to have this code written once and shared.
For now I suggest you just do whatever works best for (i.e., put
it in your bundle and make it work however is best for you) and we can
look at consolidation later if needed.
Make sense?
Jeff
For the delta processing steps I need to ´store´
an
IArtifactKey into a
string. This artifact key identifies the predecessor used for
patching/diffing.
I created a simple class performing this serialization and
de-serialization (comma separated list of the properties of
IArtifactKey).
What is better?
- Split this De/Serializer class and place the Serializer and the
De-serializer where they are used
- Keep it together (one class) and place in the package
org.eclipse.equinox.p2.metadata
- ..
Stefan
_______________________________________________
equinox-dev mailing list
equinox-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/equinox-dev
_______________________________________________
equinox-dev mailing list
equinox-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/equinox-dev
|