I think that would allow us to fix the multisearcher negative query
wildcard bug in none remote cases.
Anthony Urso wrote:
I am using the RMI RemoteSearchable. I don't use a central RMI
registry for the cluster (it's not particularly beneficial to do so),
so I have had none of the problems described above. It just works out
of the box.
Why not work towards RemoteSearchable instead of Searchable extending
java.rmi.Remote?
Cheers,
Anthony
PS: I would happily upgrade to the Hadoop RPC version of
RemoteSearchable that was discussed here earlier if it is moved into
contrib.
On Fri, Sep 26, 2008 at 9:12 AM, mark harwood <[EMAIL PROTECTED]> wrote:
since not many people, I think, even use the RMI stuff
I certainly binned RMI in my distributed work.
It just would not reliably stop/restart cleanly in my experience - despite
following all the RMI guidelines for clean shutdowns.
I'd happily see all RMI dependencies banished from core.
Cheers
Mark
----- Original Message ----
From: Grant Ingersoll <[EMAIL PROTECTED]>
To: java-dev@lucene.apache.org
Sent: Friday, 26 September, 2008 17:01:25
Subject: Re: RMI, Searchable and RemoteSearchable
Right, I think we could mark the RMI stuff on Searchable as deprecated
and note that it will be moved in 3.0.
I will open a JIRA.
On Sep 26, 2008, at 11:57 AM, DM Smith wrote:
Grant Ingersoll wrote:
Came across:
http://www.google.com/url?sa=t&source=web&ct=res&cd=1&url=http%3A%2F%2Fgroups.google.com%2Fgroup%2Fandroid-developers%2Fbrowse_thread%2Fthread%2F601329551a87e601%2Fcd0919ce891b4a26%3Flnk%3Dgst%26q%3Dlucene&ei=zNzcSPHCF4yI1ga61YiTBA&usg=AFQjCNECrBnNPBkxI4I0EbIzI-mNp3S4YA&sig2=4sDMmwa9Mrk38Nla_ULtOg
The gist of it is, that on Android doesn't provide the RMI
package. Thus, Lucene core can't work on Android b/c of Searchable
extending java.rmi.Remote.
This has always seemed a little weird to me. I'm no RMI expert,
but couldn't we drop the extension on Searchable, and add it to
RemoteSearchable, and then move RemoteSearchable and all the RMI
stuff to a separate contrib? This, in my view, would be consistent
with keeping core, core, since not many people, I think, even use
the RMI stuff and it certainly isn't required in most cases.
I haven't looked into what would break in doing it, so it is just a
suggestion at this point. Seems like it could be "mostly" back-
compatible, though, since we would just be requiring someone to get
the new JAR, unless of course they implemented their own extension
of Searchable that relied on the Remote marker interface.
Thoughts?
If done, it should be done soon. Once we get to 2.9/3.0 we are kind
of stuck for a long time.
-- DM
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
