Steve Lawrence created DAFFODIL-2888:
----------------------------------------
Summary: Incorrect out of range warning with constant dfdl:length
and length/minLength/maxLength facets
Key: DAFFODIL-2888
URL: https://issues.apache.org/jira/browse/DAFFODIL-2888
Project: Daffodil
Issue Type: Bug
Components: Front End
Reporter: Steve Lawrence
Say we have a schema like this:
{code:xml}
<element name="file" dfdl:lengthKind="explicit" dfdl:length="16"
dfdl:lengthUnits="bits">
<simpleType>
<restriction base="xs:string">
<maxLength value="2" />
</restriction>
</simpleType>
</element>
{code}
So we have a 2-byte constant length string and we specify maxLength should be
2. However, because we specify the length in bits, the value of dfdl:length is
greater than the value of maxLength and so we output a warning:
{quote}
[warn] Schema Definition Warning: Explicit dfdl:length of 16 is out of range
for facet maxLength='2'. (id: facetExplicitLengthOutOfRange)
{quote}
The issue is this check does not take into account the units of length. It also
does not take into account the character set of the data. For example, 1-byte
of the X-DFDL-BITS-MSBF character set has a minLength of 8 characters in the
infoset.
There are other things to consider, for example, maybe we have a string with
dfdl:length="3" with units of bytes, but we always expect to have 1 character
of padding removed, so our maxLength wants to be 2. We would get a warning here
too.
Maybe some false positives are okay, but we can probably improve the logic to
avoid false positives in some cases at least to take into account units and
charsets? Or maybe this warning isn't really that helpful, and the compelxity
needed to get it right is not worth it, and should be removed?
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
