uima-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Marshall Schor (JIRA)" <...@uima.apache.org>
Subject [jira] [Commented] (UIMA-5135) UIMA CasIOUtils enhancements in handling type systems
Date Fri, 07 Oct 2016 18:22:20 GMT

    [ https://issues.apache.org/jira/browse/UIMA-5135?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15555849#comment-15555849
] 

Marshall Schor commented on UIMA-5135:
--------------------------------------

Re: using Compressed forms for logging: one of the main use cases for logging is debugging.
 Writing in Xmi / Xml (for typesystem) has the advantage that no "uima" code need be used
to look at these (it's more or less directly humanly readable), and the disadvantage of being
larger/slower.  To me, this feels like a good tradeoff for debugging scenarios.

I'm thinking we should wait on implementing something here until we get a clearer sense of
what the user community may want.

> UIMA CasIOUtils enhancements in handling type systems
> -----------------------------------------------------
>
>                 Key: UIMA-5135
>                 URL: https://issues.apache.org/jira/browse/UIMA-5135
>             Project: UIMA
>          Issue Type: Improvement
>          Components: Core Java Framework
>            Reporter: Marshall Schor
>            Priority: Minor
>             Fix For: 3.0.0SDKexp, 2.9.1SDK
>
>
> A recent Jira UIMA-5120 was logging CASs to file system directories, and including a
type system.  
> It would be good to have a conventional,supported way to do this common kind of operation,
added to CasIOUtils.
> Additionally, it would be good to support as an alternative the standard XML serialization
format for type systems.
> Some possible conventions:  
> * multiple cas files, in 1 directory, with one additional file with the name "typesystem.xml".
> * the above style, in one zip file (for example, to be able to read it, one cas at a
time, via some iterator).
> * finding a type system via the class path following uimaFIT conventions
> One factor that probably is important is to store the type system for this kind of thing
"close to" the serialized forms it applies to.
> It would be possible of course to support multiple conventions.  However, the more conventions,
the less benefit from "standardization", so this ought to be a balance.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message