[ 
https://issues.apache.org/jira/browse/DAFFODIL-1639?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16841592#comment-16841592
 ] 

Michael Beckerle commented on DAFFODIL-1639:
--------------------------------------------

Pushed updates that run this under IBM cross test rig, to insure it works there 
at least.

Both tests fail under daffodil, even on the daffodil-1080-sep branch where TLog 
works.

I did have to change one expected result set to have one more element in it 
that IBM DFDL now creates.

There are two tests here now. Added the second one.

Daffodil schema-compile time is out of control for this schema. We could 
comment out a bunch of the schema to improve this, but really we need to get on 
with fixing this exponential/high-order-polynomial mistake in Daffodil schema 
compiliation.

> IBM Compatibility - EDIFACT
> ---------------------------
>
>                 Key: DAFFODIL-1639
>                 URL: https://issues.apache.org/jira/browse/DAFFODIL-1639
>             Project: Daffodil
>          Issue Type: New Feature
>          Components: Compatibility, DFDL Schemas, Integrations
>    Affects Versions: 2.0.0
>            Reporter: Michael Beckerle
>            Assignee: Michael Beckerle
>            Priority: Major
>              Labels: ForInteroperabilityTest
>             Fix For: 2.4.0
>
>
> DFDLSchemas EDIFACT doesn't yet work.
> As of this writing these issues are blocking it, or blocking working on it 
> sufficiently that they have to be fixed first.
> DFDL-1443 - escapeScheme expressions evaluated before we know there are any 
> delimiters to escape
> DFDL-256 - ignoreCase - this is used by edifact, unclear if it is truly 
> needed for the tests to work or not.
> DFDL-1166 - TDML runner needs to load a saved parser - it takes a very long 
> time to compile this schema (many minutes).  DFDL-1444 is the ticket for 
> separate compilation, but just compiling with less memory footprint and 
> faster would help.
> Currently compiling the EDIFACT schema requires sbt -mem 8000m 
> Otherwise it dies a long death as it runs out of memory, and all threads are 
> busy scavenging it.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to