Hi list,
I've encountered some strange behaviour (bug?) in Drools (4.0.7) when
adding packages from XML. Take the following rule as an example:
<rule name="constraintTest">
<lhs>
<pattern identifier="$foo" object-type="Bar" >
<field-constraint field-name="a">
<literal-restriction evaluator="==" value="A" />
</field-constraint>
<or-constraint-connective>
<and-constraint-connective>
<field-constraint field-name="b">
<literal-restriction evaluator="==" value="B"
/>
</field-constraint>
<field-constraint field-name="c">
<literal-restriction evaluator="==" value="C"
/>
</field-constraint>
</and-constraint-connective>
<field-constraint field-name="d">
<literal-restriction evaluator="==" value="D"
/>
</field-constraint>
</or-constraint-connective>
</pattern>
</lhs>
<rhs>
log("constraintTest");
</rhs>
</rule>
Trying to add a package from this gives the following Error:
org.drools.compiler.DroolsParserException:
org.xml.sax.SAXParseException: <or-constraint-connective> is after an
invalid element: org.drools.xml.Handler
at
org.drools.compiler.PackageBuilder.addPackageFromXml(PackageBuilder.java:184)
Now here's what's strange: If you rewrite the pattern like this
<pattern identifier="$foo" object-type="Bar" >
<or-constraint-connective>
<and-constraint-connective>
<field-constraint field-name="b">
<literal-restriction evaluator="==" value="B" />
</field-constraint>
<field-constraint field-name="c">
<literal-restriction evaluator="==" value="C" />
</field-constraint>
</and-constraint-connective>
<field-constraint field-name="d">
<literal-restriction evaluator="==" value="D" />
</field-constraint>
</or-constraint-connective>
<field-constraint field-name="a">
<literal-restriction evaluator="==" value="A" />
</field-constraint>
</pattern>
in other words (and I tried this with a couple of variations), if you
put all field-constraints after any and-constraint-connectives or
or-constraint-connectives sharing the same parent element, the package
adds just fine.
I couldn't find any reason, why this should be intended behaviour, so I
assume it's a bug.
I took a little peek into Drools' source, and my guess is, that there's
just some 'validPeers' missing in one of the *Handler classes that are
responsible for the validation of Drools XML configuration files.
However, this really is just my first guess.
'Adds just fine' is not the whole truth though: Modifying the pattern as
described above, still leaves me with the following message (strangely
printed to stdout, instead of using a logger or throwing an exception):
(null: 33, 46): cvc-complex-type.2.4.a: Invalid content starting with
element 'field-constraint'. One of
'{"http://drools.org/drools-4.0":and-constraint-connective}' is
expected.
Now, if I wrap the "d" field-constraint (see my example) into an
and-constraint-connective (and thus being the field-constraint the only
element in the and-constraint-connective, which looks rather useless),
that message goes away.
Even more oddness: the "a" field-constraint doesn't cause such a message
- even without a wrapping it into an and-constraint-connective.
Many thanks in advance,
Paul
Show replies by date
Hi,
this is a repost, and I'm sorry if I annoy anybody with it, but it's
already a week and nobody replied yet (to what I think might be a
serious issue for anybody using XML for configuration).
Or might this be an issue for the developers list rather than this list
and should i post it there?
So here's again, what I have problems with:
I've encountered some strange behaviour (bug?) in Drools (4.0.7) when
adding packages from XML. Take the following rule as an example:
<rule name="constraintTest">
<lhs>
<pattern identifier="$foo" object-type="Bar" >
<field-constraint field-name="a">
<literal-restriction evaluator="==" value="A" />
</field-constraint>
<or-constraint-connective>
<and-constraint-connective>
<field-constraint field-name="b">
<literal-restriction evaluator="==" value="B"
/>
</field-constraint>
<field-constraint field-name="c">
<literal-restriction evaluator="==" value="C"
/>
</field-constraint>
</and-constraint-connective>
<field-constraint field-name="d">
<literal-restriction evaluator="==" value="D"
/>
</field-constraint>
</or-constraint-connective>
</pattern>
</lhs>
<rhs>
log("constraintTest");
</rhs>
</rule>
Trying to add a package from this gives the following Error:
org.drools.compiler.DroolsParserException:
org.xml.sax.SAXParseException: <or-constraint-connective> is after an
invalid element: org.drools.xml.Handler
at
org.drools.compiler.PackageBuilder.addPackageFromXml(PackageBuilder.java:184)
Now here's what's strange: If you rewrite the pattern like this
<pattern identifier="$foo" object-type="Bar" >
<or-constraint-connective>
<and-constraint-connective>
<field-constraint field-name="b">
<literal-restriction evaluator="==" value="B" />
</field-constraint>
<field-constraint field-name="c">
<literal-restriction evaluator="==" value="C" />
</field-constraint>
</and-constraint-connective>
<field-constraint field-name="d">
<literal-restriction evaluator="==" value="D" />
</field-constraint>
</or-constraint-connective>
<field-constraint field-name="a">
<literal-restriction evaluator="==" value="A" />
</field-constraint>
</pattern>
in other words (and I tried this with a couple of variations), if you
put all field-constraints after any and-constraint-connectives or
or-constraint-connectives sharing the same parent element, the package
adds just fine.
I couldn't find any reason, why this should be intended behaviour, so I
assume it's a bug.
I took a little peek into Drools' source, and my guess is, that there's
just some 'validPeers' missing in one of the *Handler classes that are
responsible for the validation of Drools XML configuration files.
However, this really is just my first guess.
'Adds just fine' is not the whole truth though: Modifying the pattern as
described above, still leaves me with the following message (strangely
printed to stdout, instead of using a logger or throwing an exception):
(null: 33, 46): cvc-complex-type.2.4.a: Invalid content starting with
element 'field-constraint'. One of
'{"http://drools.org/drools-4.0":and-constraint-connective}' is
expected.
Now, if I wrap the "d" field-constraint (see my example) into an
and-constraint-connective (and thus being the field-constraint the only
element in the and-constraint-connective, which looks rather useless),
that message goes away.
Even more oddness: the "a" field-constraint doesn't cause such a message
- even without a wrapping it into an and-constraint-connective.
Many thanks in advance,
Paul