uima-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eddie Epstein <eaepst...@gmail.com>
Subject Re: small memory footprint tradeoff configuration
Date Mon, 23 Feb 2009 20:38:20 GMT
On Mon, Feb 23, 2009 at 12:10 PM, Thilo Goetz <twgoetz@gmx.de> wrote:
> So the users are supposed to figure out if they need internal
> IDs?  I don't think that's a good idea.  Either we make guarantees
> about things like references into the CAS surviving calls to
> process(), or we don't.

Process calls to a Vinci service have always broken FS references.
Same for calls thru the compatibility wrapper that allows calling
colocated UIMA 1.4x annotators from Apache UIMA. So we have not in the
past made such guarantees for remote or even all colocated components.

Supporting GC called in a service will require some work. If the
client uses XMI serialization and supports delta CAS replies, nothing
would change and the GC would be a noop as far as the client is
concerned. With the entire CAS coming back, some changes would be
needed to compensate for deletion of pre-existing FS, but otherwise FS
references would still be good. A binary serialization reply after GC
would definitely invalidate references.

Eddie

Mime
View raw message