[
https://issues.apache.org/jira/browse/DAFFODIL-2703?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17552783#comment-17552783
]
Mike Beckerle commented on DAFFODIL-2703:
-----------------------------------------
If fillByte %r00; unparses properly, and fillByte %rFF; unparses properly, and
they both agree for runtime 1 and runtime 2, then the priority is lower since
those are the two cases people will actually use.
If the above holds, then I think a release not suggesting only using 0x00 or
0xFF as fillByte would suffice for a long time.
Otherwise this bug is pretty high priority to fix, because this will make test
data and TDML tests much harder to deal with, requiring different tests for the
two runtimes where otherwise they could be shared. Tests may also not round
trip where they otherwise should.
> Unaligned RightFill does not unparse correctly
> ----------------------------------------------
>
> Key: DAFFODIL-2703
> URL: https://issues.apache.org/jira/browse/DAFFODIL-2703
> Project: Daffodil
> Issue Type: Bug
> Components: Back End, Back End C-Generator
> Affects Versions: 3.3.0
> Reporter: Steve Lawrence
> Priority: Major
> Attachments: unalignedRightFill.tdml
>
>
> See the attached unalignedRightFill.tdml file for reproducible tests.
> Data is a single byte. Only the most significant bit is actual data. The
> remaining 7 bits are RightFill region using a fillByte value of 0xF0. So the
> data should always unparse to X111 0000, where the value of X comes from the
> infoset.
> Note that runtime1 and runtime2 have different behaviors, neither of which I
> think is correct:
>
>
> *unalignedRightFill_01* (bit from infoset is 1)
> should unparse to 0xF0 (1111 0000)
> runtime1 unparses to 0xBF (1011 1111)
> runtime2 unparses to 0x9F (1001 1111)
>
> *unalignedRightFill_02* (bit from infoset is 0)
> should unparse to 0x70 (0111 0000)
> runtime1 unaprses to 0x3F (0011 1111)
> runtime2 unparses to 0x1F (0001 1111)
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
