Saurabh Vyas wrote:
Brock Pytlik wrote:
One more thought, the right place to fix this is in attribute.py. I
think if 'None' on line 109 was change to 'v', the correct behavior
would happen. The problem with just doing this is that it requires a
reindex to work correctly, which essentially means it means bumping
the dictionary format.
I will try this approach too, but I don't know how to initialize
re-indexing
Take a look at the indexing code and the search_storage code, but the
short version is that changing the version on the file causes a reindex.
(It would also force a reindex on the client, but that's probably
less of an issue. We could also discuss ways to make it work
compatibly with the old version until the user does a full-reindex
automatically.) Further (without a corresponding bump in the search
version), a client wouldn't necessarily know whether it was talking
to a server which was working from an old or new version. However, we
could do a decent job making a guess (and it wouldn't be any more
gross than what we'd need to do to handle the multiple value action
case anyway).
Please excuse me in this but I did not understood this completely.
Does this mean that the version of behavior can change the result and
we need to handle this on client side (by making proper assumptions).
I don't understand your question here. I'm not certain what "version of
behavior" is referring to.
In short, for local search, the client controls the index and the
processing, so it's not as big of an issue there unless we decide not to
force a reindex. However for remote search, unless we bump the search
version, the client has no way of knowing whether it's talking to a
server with this fix on it or not. What the server sends to the client
will differ depending on whether the fix is there or not, so the client
needs to appropriately handle both cases.
Brock
Thanks,
~Saurabh
Brock
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss