[
https://issues.apache.org/jira/browse/DAFFODIL-1898?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17420160#comment-17420160
]
Mike Beckerle edited comment on DAFFODIL-1898 at 9/25/21, 8:20 PM:
-------------------------------------------------------------------
I found this comment in the TDMLRunner in the definition of method
{code:java}
def decodeDataToString(decoder: BitsCharsetDecoder, bytes: Array[Byte]): String
....
// Decoders don't actually need much of the formatinfo. Byte size
// decoders don't ever refernce byte/bitOrder, they just read raw bytes
// from the underlying data stream. And non-byte size encoders do care,
// but only support bigEndian/LSBF. There should never be encoding
// errors, so the error policy doesn't really matter.
{code}
This is patently false at this point, there are not byte-sized encodings that
use both MSBF and LSBF bit order.
The code does seem to ignore the bitOrder of the encoding itself and just
always use BE, LSBF.
This seems likely to be the cause of the problem.
was (Author: mbeckerle):
I found this comment in the TDMLRunner in the definition of method
{code:java}
def decodeDataToString(decoder: BitsCharsetDecoder, bytes: Array[Byte]): String
....
// Decoders don't actually need much of the formatinfo. Byte size
// decoders don't ever refernce byte/bitOrder, they just read raw bytes
// from the underlying data stream. And non-byte size encoders do care,
// but only support bigEndian/LSBF. There should never be encoding
// errors, so the error policy doesn't really matter.
{code}
This is patently false at this point, there are not byte-sized encodings that
use both MSBF and LSBF bit order.
This seems likely to be the cause of the problem.
> TDML Runner not assembling mixtures of BE+MSBF and LE+LSBF test data properly.
> ------------------------------------------------------------------------------
>
> Key: DAFFODIL-1898
> URL: https://issues.apache.org/jira/browse/DAFFODIL-1898
> Project: Daffodil
> Issue Type: Bug
> Components: TDML Runner
> Affects Versions: 2.0.0
> Reporter: Mike Beckerle
> Priority: Major
>
> See test testMixedBigEndianMSBFWithLittleEndianLSBF
> Also the test test_ep3 had to work around the bug. See the tdml file for this
> test and see the comments and work-around. This test ought to be changed to
> not use the workaround.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)