mbeckerle commented on a change in pull request #259: Incremental progress on schema compilation space/speed issue. URL: https://github.com/apache/incubator-daffodil/pull/259#discussion_r303910031
########## File path: daffodil-test/src/test/resources/org/apache/daffodil/section23/dfdl_expressions/expressions3.tdml ########## @@ -0,0 +1,400 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!-- + Licensed to the Apache Software Foundation (ASF) under one or more + contributor license agreements. See the NOTICE file distributed with + this work for additional information regarding copyright ownership. + The ASF licenses this file to You under the Apache License, Version 2.0 + (the "License"); you may not use this file except in compliance with + the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + + Unless required by applicable law or agreed to in writing, software + distributed under the License is distributed on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + See the License for the specific language governing permissions and + limitations under the License. +--> + +<tdml:testSuite + suiteName="expressions3" + xmlns:tdml="http://www.ibm.com/xmlns/dfdl/testData"; + xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"; + xmlns:xs="http://www.w3.org/2001/XMLSchema"; + xmlns:dfdl="http://www.ogf.org/dfdl/dfdl-1.0/"; + xmlns:dfdlx="http://www.ogf.org/dfdl/dfdl-1.0/extensions"; + xmlns:ex="http://example.com"; + xmlns:fn="http://www.w3.org/2005/xpath-functions"; + defaultRoundTrip="true"> + + <tdml:defineSchema name="s1" elementFormDefault="unqualified"> + <xs:include schemaLocation="org/apache/daffodil/xsd/DFDLGeneralFormat.dfdl.xsd" /> + <dfdl:format ref="ex:GeneralFormat" lengthKind="delimited"/> + + <xs:group name="g"> + <!-- + The path expression below is going to have different type + elements for the different instances of this group in the + final schema. + + Ultimately they all are converted to string. + --> + <xs:sequence> + <xs:choice dfdl:choiceDispatchKey="{ ../foo/bar }"> + <xs:element name="c1" type="xs:string" dfdl:choiceBranchKey="1" /> + <xs:element name="c2" type="xs:string" dfdl:choiceBranchKey="2" /> + </xs:choice> + </xs:sequence> + </xs:group> + + <xs:element name="r"> + <xs:complexType> + <xs:sequence dfdl:separator="|"> + <xs:element name="e1"> + <xs:complexType> + <xs:sequence> + <xs:element name="foo"> + <xs:complexType> + <xs:sequence dfdl:separator="|"> + <xs:element name="bar" type="xs:string" /> + <xs:group ref="ex:g" /> + </xs:sequence> + </xs:complexType> + </xs:element> + </xs:sequence> + </xs:complexType> + </xs:element> + <xs:element name="e2"> + <xs:complexType> + <xs:sequence> + <xs:element name="foo"> + <xs:complexType> + <xs:sequence dfdl:separator="|"> + <xs:element name="bar" type="xs:int" /> + <xs:group ref="ex:g" /> + </xs:sequence> + </xs:complexType> + </xs:element> + </xs:sequence> + </xs:complexType> + </xs:element> + </xs:sequence> + </xs:complexType> + </xs:element> + </tdml:defineSchema> + + <tdml:parserTestCase name="test_polymorphic_expr_1" model="s1" root="r" + description="Polymorphic path expression"> + <tdml:document>1|a|2|b</tdml:document> + <tdml:infoset> + <tdml:dfdlInfoset> + <ex:r> + <e1> + <foo> + <bar>1</bar> + <c1>a</c1> + </foo> + </e1> + <e2> + <foo> + <bar>2</bar> + <c2>b</c2> + </foo> + </e2> + </ex:r> + </tdml:dfdlInfoset> + </tdml:infoset> + <tdml:warnings> + <tdml:warning>should be manually cast</tdml:warning> + </tdml:warnings> + </tdml:parserTestCase> + + <tdml:defineSchema name="s2a" elementFormDefault="unqualified"> Review comment: Yes. I guess I agree. Full referential transparency should be the goal. Whether you copy the reusable thing inline or keep it shared should not change the behavior. There is another alternative - parametric - require people to bind variables rather than allow these upward-and-outward paths. The parameters have types. Another equivalent possibility: require encapsulating any upward-and-outward path expression in a constructor call.. So instead of ../../foo/bar you must write xs:double( ../../foo/bar ) (or other constructor) which is another way of forcing a type-declaration - in this case the type of foo/bar must be suitable to be an arg to the double constructor, which means I believe that it just has to have a value, since those constructors even take strings. This is, I guess, just weakening the type checking substantially. Ultimately, keeping track of every point of usage, the types at those points of use, and generating N specializations for the N actual situations that arise, is not impossible. It's just harder. E.g., consider an expression with multiple paths in it, each of which is upward-and-out to a variety of different types. The number of instance you need is potentially the product of those.... I would probably approximate this as follows: If there is exactly one type for all usages share. Otherwise copy. I would certainly hope the vast bulk of expressions are used monomorphically. Are there any other ideas for how to achieve full referential transparency here? ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
