Application Description Language is an XML vocabulary, defined in a Document Type Definition, which declaratively describes the entities in an application domain, their relationships, and their properties. Because ADL is defined in a formal definition which can be parsed by XML editors, any DTD-aware XML editor (such as that built into Visual studio) can provide context-sensitive auto-completion for ADL, making the vocabulary easy to learn and to edit. It would perhaps be desirable to replace this DTD at some future stage with an XML Schema, since it is desirable to be able to mix HTML in with ADL in the same document.
ADL is thus a 'Fourth Generation Language' as understood in the 1980s - an ultra-high level language for a specific problem domain; but it is a purely declarative 4GL.
The Application Description Language Framework is principally a set of XSL transforms which transform a single ADL file into all the various source files required to build an application.
The average data driven web application comprises pages (lists) which show lists of entities, pages (forms) that edit instances of entities, and pages (inspectors) that show details of instances of entities. That comprises 100% of many applications and 90% of others; traditionally, even with modern tools like Monorail, coding these lists, forms and inspectors has taken 90% of the development effort.
I realised about three years ago that I was doing essentially the same job over and over again, and I don't like doing that. I see my mission in life as being to automate people out of jobs, and that includes me. So the object of the Application Description Language is to raise the level of abstraction with which we define data driven applications one level higher, and automate the process we have thus far done as programmers. This isn't a new insight; it's fundamentally the same insight that led machine code programmers to develop the first macro assembler, and led assembly language programmers to write the first high level language compiler. Computers are tools which can be used to mung information from one representation to another, and all we need to do is to work out how to write a powerful enough representation, and how to transform it.
The whole purpose of ADL is to increase productivity - mine, and that of anyone else who chooses to follow me down this path. It is pragmatic technology - it is designed to be an 80/20 or 90/10 solution, taking the repetitious grunt-work out of application development so that we can devote more time to the fun, interesting and novel bits. It is not intended to be an academic, perfect, 100% solution - although for many applications it may in practice be a 100% solution.
Thus to create a new application, all that should be necessary is to create a new ADL file, and to compile it using a single, standardised [NAnt] (or [Ant]) build file using scripts already created as part of the framework. All these scripts (with the exception of the PSQL one, which was pre-existing) have been created as part of the C1873 - SRU - Hospitality contract, but they contain almost no SRU specific material (and what does exist has been designed to be factored out). Prototype 1 of the SRU Hospitality Application contains no hand-written code whatever - all the application code is automatically generated from the single ADL file. The one exception to this rule is the CSS stylesheet which provides look-and-feel and branding.
Application-specific procedural code, covering specific business procedures, may still need to be hand written; the code generated by the ADL framework is specifically designed to make it easy to integrate hand-written code. Thus for example the C# entity controller classes generated are intentionally generated as partial classes, so that they may be complemented by other partial classes which may be manually maintained and held in a version control system.
One key objective of the framework is that the code which is generated should be as clear and readable - and as well commented - as the best hand-written code. Consider this example:
/// <summary>
/// Store the record represented by the parameters passed in an HTTP service
/// Without Id -> it's new, I create a new persistent object;
/// With Id -> it's existing, I update the existing persistent object
/// </summary>
[AccessibleThrough( Verb.Post)]
public void Store()
{
ISession hibernator =
NHibernateHelper.GetCurrentSession( Session[ NHibernateHelper.USERTOKEN],
Session[NHibernateHelper.PASSTOKEN]);
SRU.Hospitality.Entities.Event record;
if ( Params[ "instance.Date" ] == null)
{
AddError( "You must supply a value for Date");
}
if ( Params[ "instance.Description" ] == null)
{
AddError( "You must supply a value for Description");
}
string id = Params["instance.EventId"];
if ( String.IsNullOrEmpty( id))
{
/* it's new, create persistent object */
record = new SRU.Hospitality.Entities.Event();
/* perform any domain knowledge behaviour on the new record
* after instantiation */
record.AfterCreationHook();
}
else
{
/* it's existing, retrieve it */
record =
hibernator.CreateCriteria(typeof(Event))
.Add(Expression.Eq("EventId", Int32.Parse(id)))
.UniqueResult<SRU.Hospitality.Entities.Event>();
}
if ( record != null)
{
/* perform any domain knowledge behaviour on the record prior to updating */
record.BeforeUpdateHook();
/* actually update the record */
BindObjectInstance( record, ParamStore.Form, "instance");
/* write the record to the database, in order to guarantee we have a valid key */
hibernator.Save(record);
hibernator.Flush();
/* perform any domain knowledge behaviour on the record after updating */
record.AfterUpdateHook();
PropertyBag["username"] = Session[ NHibernateHelper.USERTOKEN];
PropertyBag["instance"] = record;
RenderViewWithFailover("edit.vm", "edit.auto.vm");
}
else
{
throw new Exception( String.Format( "No record of type Event with key value {0} found", id));
}
}
This means that it should be trivial to decide at some point in development of a project to manually modify and maintain auto-generated code.
Currently the framework includes:
Transforms the ADL file into C# source files for classes which describe the entities in a manner acceptable to NHibernate, a widely used Object/Relational mapping layer.
Transforms the ADL file into an SQL script in Microsoft SQL Server 2000 syntax which initialises the database required by the application, with all relationships, permissions, referential integrity constraints and so on.
Transforms the ADL file into Velocity template files as used by the Monorail framework, one template each for all the lists, forms and inspectors described in the ADL.
Transforms the ADL file into a series of C# source files for classes which are controllers as used by the Monorail framework.
Transforms the ADL file into a Hibernate mapping file, used by the Hibernate (Java) and NHibernate (C#) Object/Relational mapping layers. This transform is relatively trivial, since ADL is not greatly different from being a superset of the Hibernate vocabulary - it describes the same sorts of things but in more detail.
Transforms the ADL file into an SQL script in Postgres 7 syntax which initialises the database required by the application, with all relationships, permissions, referential integrity constraints and so on.
Yes and no.
ADL is a quick way to build Monorail applications, because it seemed to me that as Monorail/NHibernate are technologies that the company is adopting and it would be better to work with technologies with which we already have expertise - it's no good doing these things if other people can't maintain them afterwards.
However ADL wasn't originally conceived with Monorail in mind. It was originally intended to generated LISP for CLHTTPD, and I have a half-finished set of scripts to generate Java as part of the Jacquard2 project which I never finished. Because ADL is at a level of abstraction considerably above any 3GL, it is inherently agnostic to what 3GL it is compiled down to - so that it would be as easy to write transforms that compiled ADL to Struts or Ruby on Rails as to C#/Monorail. More importantly, ADL isn't inherently limited to Web applications - it doesn't actually know anything about the Web. It should be possible to write transforms which compile ADL down to Windows native applications or to native applications for mobile phones (and, indeed, if we did have those transforms then we could make all our applications platform agnostic).
Although I've built experimental systems before using ADL, the SRU project is the first time I've really used it in anger. There are some features I need which it can't yet represent.
For SRU, I have implemented an authentication model which authenticates the user against real database user accounts. I've done this because I think, in general, this is the correct solution, and because without this sort of authentication you cannot implement table-layer security. However most web applications use application layer authentication rather than database layer authentication, and I have not yet written controller-layer code to deal with this. So unless you do so, ADL applications can currently only authenticate at database layer.
ADL defines field-level permissions, but the current controller generator does not implement this.
Generically, with an entity form, one needs to be able to save the record being edited, and one (often) needs to be able to delete it. But sometimes one needs to be able to do other things. With SRU, for example, there is a need to be able to export event data to Perfect Table Plan, and to reimport data from Perfect Table Plan. This will need custom buttons on the event entity form, and will also need hand-written code at the controller layer to respond to those buttons.
Also, a person will have, over the course of their interaction with the SRU, potentially many invitations. In order to access those invitations it will be necessary to associate lists of dependent records with forms. Currently ADL cannot represent these.
At this stage I doubt whether there is much point in extending ADL to include a vocabulary to describe business processes. It would make the language much more complicated, and would be unlikely to be able to offer a significantly higher level of abstraction than current 3GLs. If using ADL does not save work, it isn't worth doing it in ADL; remember this is conceived as an 80/20 solution, and you need to be prepared to write the 20 in something else.
This section of this document presents and comments on the existing ADL document type definition (DTD).
The DTD starts with some basic definitions
<!-- :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: -->
<!-- Before we start: some useful definitions -->
<!-- :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: -->
<!-- boolean means true or false -->
<!ENTITY % Boolean "(true|false)" >
<!--
Locale is a string comprising an ISO 639 language code followed by a space
followed by an ISO 3166 country code, or else the string 'default'. See:
<URL:http://www.ics.uci.edu/pub/ietf/http/related/iso639.txt>
<URL:http://www.chemie.fu-berlin.de/diverse/doc/ISO_3166.html>
-->
<!ENTITY % Locale "CDATA" >
Key to any data driven application is who has authority to do what to what: 'permissions'.
<!--
permissions a group may have on an entity, list, page, form or field
permissions are deemed to increase as you go right. A group cannot
have greater permission on a field than on the form it is in, or
greater permission on form than the entity it belongs to
none: none
read: select
insert: insert
noedit: select, insert
edit: select, insert, update
all: select, insert, update, delete
-->
<!ENTITY % Permissions "none|read|insert|noedit|edit|all" >
ADL needs to know what type of data can be stored on different properties of different entities. The data types were originally based on JDBC data types:
<!--
data types which can be used in a definition to provide validation -
e.g. a string can be used with a regexp or a scalar can be used with
min and max values
string: varchar java.sql.Types.VARCHAR
integer: int java.sql.Types.INTEGER
real: double java.sql.Types.DOUBLE
money: money java.sql.Types.INTEGER
date: date java.sql.Types.DATE
time: time java.sql.Types.TIME
timestamp: timestamp java.sql.Types.TIMESTAMP
-->
However, in order to be able to do data validation, it's useful to associate rules with data types. ADL has the concept of definable data types, to allow data validation code to be generated from the declarative description. These definable data types are used in the ADL application, for example, to define derived types for phone numbers, email addresses, postcodes, and range types.
<!ENTITY % DefinableDataTypes "string|integer|real|money|date|time|timestamp" >
<!--
data types which are fairly straightforward translations of JDBC data types
boolean: boolean or java.sql.Types.BIT
char(1) java.sql.Types.CHAR
text: text or java.sql.Types.LONGVARCHAR
memo java.sql.Types.CLOB
-->
<!ENTITY % SimpleDataTypes "%DefinableDataTypes;|boolean|text" >
<!--
data types which are more complex than SimpleDataTypes...
entity : a foreign key link to another entity;
link : a many to many link (via a link table);
defined : a type defined by a definition.
-->
<!ENTITY % ComplexDataTypes "entity|link|defined" >
<!-- all data types -->
<!ENTITY % AllDataTypes "%ComplexDataTypes;|%SimpleDataTypes;" >
Pages in applications typically have common, often largely
static, sections above, below, to the left or right of the main
content which incorporates things like branding, navigation,
and so on. This can be defined globally or per page. The
intention is that the head, top and
foot elements in ADL should be allowed to contain
arbitrary HTML, but currently I don't have enough skill with
DTD design to know how to specify this.
<!-- content, for things like pages (i.e. forms, lists, pages) -->
<!ENTITY % Content "head|top|foot" >
<!ENTITY % PageContent "%Content;|field" >
<!ENTITY % PageStuff "%PageContent;|permission|pragma" >
<!ENTITY % PageAttrs
"name CDATA #REQUIRED
properties (all|listed) #REQUIRED" >
The top level element of an Application Description Language file is the application element:
<!-- the application that the document describes: required top level element -->
<!ELEMENT application ( content?, definition*, group*, entity*)>
<!ATTLIST application
name CDATA #REQUIRED
version CDATA #IMPLIED>
In order to be able to use defined types, you need to be able to provide definitions of these types:
<!--
the definition of a defined type. At this stage a defined type is either
a string in which case it must have size and pattern, or
a scalar in which case it must have minimum and/or maximum
pattern must be a regular expression as interpreted by org.apache.regexp.RE
minimum and maximum must be of appropriate format for the datatype specified.
Validation may be done client-side and/or server-side at application layer
and/or server side at database layer.
-->
<!ELEMENT definition (help*) >
<!ATTLIST definition
name CDATA #REQUIRED
type (%DefinableDataTypes;) #REQUIRED
size CDATA #IMPLIED
pattern CDATA #IMPLIED
minimum CDATA #IMPLIED
maximum CDATA #IMPLIED>
In order to be able to user permissions, we need to define who has those permissions. Groups in ADL map directly onto groups/roles at SQL level, but the intention with ADL is that groups should be defined hierarchically.
<!-- a group of people with similar permissions to one another --> <!ELEMENT group EMPTY> <!-- the name of this group --> <!ATTLIST group name CDATA #REQUIRED> <!-- the name of a group of which this group is subset --> <!ATTLIST group parent CDATA #IMPLIED>
A thing-in-the-domain has properties. Things in the domain fall into regularities, groups of things which share similar collections of properties, such that the values of these properties may have are constrained. This is a representation of the world which is not perfect, but which is sufficiently useful to be recognised by the software technologies which ADL abstracts, so we need to be able to define these. Hence we have entities and properties/
<!--
an entity which has properties and relationships; maps onto a database
table or a Java serialisable class - or, of course, various other things
-->
<!ELEMENT entity ( content?, property*, permission*, (form | page | list)*)>
<!ATTLIST entity name CDATA #REQUIRED>
<!--
a property (field) of an entity (table)
name: the name of this property.
type: the type of this property.
default: the default value of this property. There will probably be
magic values of this!
definition: name of the definition to use, it type = 'defined'.
distinct: distinct='system' required that every value in the system
will be distinct (i.e. natural primary key);
distinct='user' implies that the value may be used by users
in distinguishing entities even if values are not formally
unique;
distinct='all' implies that the values are formally unique
/and/ are user friendly.
entity: if type='entity', the name of the entity this property is
a foreign key link to.
required: whether this propery is required (i.e. 'not null').
size: fieldwidth of the property if specified.
-->
<!ELEMENT property ( option*, prompt*, help*, ifmissing*)>
<!ATTLIST property
name CDATA #REQUIRED
type (%AllDataTypes;) #REQUIRED
default CDATA #IMPLIED
definition CDATA #IMPLIED
distinct (none|all|user|system) #IMPLIED
entity CDATA #IMPLIED
required %Boolean; #IMPLIED
size CDATA #IMPLIED>
Sometimes a property has a constrained list of specific values; this is represented for example in the enumerated types supported by many programming languages. Again, we need to be able to represent this.
<!--
one of an explicit list of optional values a property may have
NOTE: whether options get encoded at application layer or at database layer
is UNDEFINED; either behaviour is correct. If at database layer it's also
UNDEFINED whether they're encoded as a single reference data table or as
separate reference data tables for each property.
-->
<!ELEMENT option (prompt*)>
<!-- if the value is different from the prompt the user sees, specify it -->
<!ATTLIST option value CDATA #IMPLIED>
Permissions define policies to allow groups of users to access forms, pages, fields (not yet implemented) or entities. Only entity permissions are enforced at database layer, and field protection is not yet implemented at controller layer. But the ADL allows it to be described, and future implementations of the controller generating transform will do this.
<!--
permissions policy on an entity, a page, form, list or field
group: the group to which permission is granted
permission: the permission which is granted to that group
-->
<!ELEMENT permission EMPTY>
<!ATTLIST permission
group CDATA #REQUIRED
permission (%Permissions;) #REQUIRED>
Pragmas are currently not used at all. They are there as a possible means to provide additional controls on forms, but may not be the correct solutions for that.
<!--
pragmatic advice to generators of lists and forms, in the form of
name/value pairs which may contain anything. Over time some pragmas
will become 'well known', but the whole point of having a pragma
architecture is that it is extensible.
-->
<!ELEMENT pragma EMPTY>
<!ATTLIST pragma
name CDATA #REQUIRED
value CDATA #REQUIRED>
When soliciting a value for a property from the user, we need to be able to offer the user a prompt to describe what we're asking for, and we need to be able to offer that in the user's preferred natural language. Prompts are typically brief. Sometimes, however, we need to give the user a more extensive description of what is being solicited - 'help text'. Finally, if the data offered by the user isn't adequate for some reason, we need ways of feeding that back. Currently the only error text which is carried in the ADL is 'ifmissing', text to be shown if the value for a required property is missing. All prompts, helptexts and error texts have locale information, so that it should be possible to generate variants of all pages for different natural languages from the same ADL.
<!--
a prompt for a property or field; used as the prompt text for a widget
which edits it. Typically there will be only one of these per property
per locale; if there are more than one all those matching the locale may
be concatenated, or just one may be used.
prompt: the prompt to use
locale: the locale in which to prefer this prompt
-->
<!ELEMENT prompt EMPTY>
<!ATTLIST prompt
prompt CDATA #REQUIRED
locale %Locale; #IMPLIED >
<!--
helptext about a property of an entity, or a field of a page, form or
list, or a definition. Typically there will be only one of these per property
per locale; if there are more than one all those matching the locale may
be concatenated, or just one may be used.
locale: the locale in which to prefer this prompt
-->
<!ELEMENT help (#PCDATA)>
<!ATTLIST help
locale %Locale; #IMPLIED >
<!--
helpful text to be shown if a property value is missing, typically when
a form is submitted. Typically there will be only one of these per property
per locale; if there are more than one all those matching the locale may
be concatenated, or just one may be used. Later there may be more sophisticated
behaviour here.
-->
<!ELEMENT ifmissing (#PCDATA)>
<!ATTLIST ifmissing
locale %Locale; #IMPLIED>
The basic pages of the user interface. Pages and Forms by default show fields for all the properties of the entity they describe, or they may show only a listed subset. Currently lists show fields for only those properties which are 'user distinct'. Forms, pages and lists may each have their own head, top and foot content, or they may inherit the content defined for the application.
<!-- a form through which an entity may be added or edited -->
<!ELEMENT form ( %PageStuff;)*>
<!ATTLIST form %PageAttrs;>
<!-- a page on which an entity may be displayed -->
<!ELEMENT page ( %PageStuff;)*>
<!ATTLIST page %PageAttrs;>
<!--
a list on which entities of a given type are listed
onselect: name of form/page/list to go to when
a selection is made from the list
-->
<!ELEMENT list ( %PageStuff;)*>
<!ATTLIST list %PageAttrs;
onselect CDATA #IMPLIED >
<!-- a field in a form or page -->
<!ELEMENT field (prompt*, help*, permission*) >
<!ATTLIST field property CDATA #REQUIRED >
<!-- a container for global content -->
<!ELEMENT content (%Content;)*>
<!--
content to place in the head of the generated document; this is #PCDATA
because it will almost certainly belong to a different namespace
(usually HTML)
-->
<!ELEMENT head (#PCDATA) >
<!--
content to place in the top of the body of the generated document;
this is #PCDATA because it will almost certainly belong to a different
namespace (usually HTML)
-->
<!ELEMENT top (#PCDATA) >
<!--
content to place at the foot of the body of the generated document;
this is #PCDATA because it will almost certainly belong to a different
namespace (usually HTML)
-->
<!ELEMENT foot (#PCDATA) >
Current versions of ADL are given at the top of this document. Historical versions are as follows:
To use ADL, it is currently most convenient to use NAnt. It is probably possible to do this with MSBuild, but as of yet I don't know how.
For the examples given here to work, you will need to set up
at least the following properties in your NAnt
.build file:
<property name="project.name" value="YourProjectName"/>
<property name="src.dir" value="YourSourceDir"/>
<property name="tmpdir" value="tmp"/>
<property name="assembly" value="${project.name}"/>
<property name="adl" value="L:/adl/unstable/ADL/"/>
<property name="adl-transforms" value="${adl}/transforms"/>
<property name="adl-src" value="${src.dir}/${project.name}.adl.xml"/>
<property name="canonical" value="${tmpdir}/Canonical.adl.xml"/>
<property name="nant-tasks" value="${tmpdir}/NantTasks.dll"/>
<property name="nsroot" value="Uk.Co.Cygnets"/>
<property name="entityns" value="${nsroot}.${assembly}.Entities"/>
<property name="controllerns" value="${nsroot}.${assembly}.Controllers"/>
<property name="entities" value="${src-dir}/Entities"/>
<property name="controllers" value="${src-dir}/Controllers"/>
where, obviously, YourProjectName, YourSourceDir and YourADL.adl.xml stand in for the actual names of your project, your source directory (relative to your solution directory, where the .build file is) and your ADL file, respectively. Note that if it is to be used as an assembly name, the project name should include neither spaces, hyphens nor periods. If it must do so, you should give an assembly name which does not, explicitly.
The first thing you need to do with your ADL file is canonicalise it. You should generally not need to alter this, you should copy and paste it verbatim:
<target name="canonicalise" description="canonicalises adl">
<style verbose="true" style="${adl-transforms}/adl2canonical.xslt"
in="${adl-src}"
out="${canonical}">
<parameters>
<parameter name="abstract-key-name-convention" value="Name_Id"/>
</parameters>
</style>
</target>
You should generally not need to alter this at all, just copy and paste it verbatim:
<target name="hbm" description="generates NHibernate mapping for database"
depends="canonicalise">
<style verbose="true" style="${adl-transforms}/adl2hibernate.xslt"
in="${canonical}"
out="${src.dir}/${project.name}.auto.hbm.xml">
<parameters>
<parameter name="namespace" value="${entityns}"/>
<parameter name="assembly" value="${assembly}"/>
</parameters>
</style>
</target>
<target name="sql" description="Generates cadlink database initialisation script"
depends="canonicalise">
<style verbose="true" style="${adl-transforms}/adl2mssql.xslt"
in="${canonical}"
out="${src.dir}/${project.name}.auto.sql">
<parameters>
<parameter name="abstract-key-name-convention" value="Name_Id"/>
<parameter name="database" value="ESA-McIntosh-CADLink"/>
</parameters>
</style>
</target>
Note that for this to work you must have the following:
c:\Program
Files\astyle\bin\astyle.exe
<target name="fetchtasks" depends="prepare"
description="fetches our NantTasks library from the well known place where it resides">
<get src="http://libs.cygnets.co.uk/NantTasks.dll"
dest="${nant-tasks}"/>
</target>
<target name="classes" description="creates C# classes for entities in the database"
depends="fetchtasks canonicalise">
<loadtasks assembly="${nant-tasks}" />
<style verbose="true" style="${adl-transforms}/adl2entityclass.xslt"
in="${canonical}"
out="${tmpdir}/classes.auto.cs">
<parameters>
<parameter name="locale" value="en-UK"/>
<parameter name="controllerns" value="${controllerns}"/>
<parameter name="entityns" value="${entityns}"/>
</parameters>
</style>
<exec program="c:\Program Files\astyle\bin\astyle.exe"
basedir="."
commandline="--style=java --indent=tab=4 --indent-namespaces ${tmpdir}/classes.auto.cs"/>
<split-regex in="${tmpdir}/classes.auto.cs"
destdir="${src.dir}/Entities"
pattern="cut here: next file '([a-zA-Z0-9_.]*)'"/>
</target>
Note that for this to work you must have
c:\Program
Files\astyle\bin\astyle.exe
<target name="controllers" description="creates C# controller classes"
depends="fetchtasks canonicalise">
<loadtasks assembly="${nant-tasks}" />
<loadtasks assembly="${nant-contrib}" />
<style verbose="true" style="${adl-transforms}/adl2controllerclasses.xslt"
in="${canonical}"
out="${tmpdir}/controllers.auto.cs">
<parameters>
<parameter name="locale" value="en-UK"/>
<parameter name="controllerns" value="${controllerns}"/>
<parameter name="entityns" value="${entityns}"/>
<parameter name="layout-name" value="default"/>
<parameter name="rescue-name" value="generalerror"/>
</parameters>
</style>
<exec program="c:\Program Files\astyle\bin\astyle.exe"
basedir="."
commandline="--style=java --indent=tab=4 --indent-namespaces ${tmpdir}/controllers.auto.cs"/>
<split-regex in="${tmpdir}/controllers.auto.cs"
destdir="${controllers}/Auto" pattern="cut here: next file '([a-zA-Z0-9_.]*)'"/>
</target>
Note that for this to work you must have
<target name="views" description="creates Velocity templates"
depends="fetchtasks canonicalise">
<loadtasks assembly="${nant-tasks}" />
<style verbose="true" style="${adl-transforms}/adl2views.xslt"
in="${canonical}"
out="${tmpdir}/views.auto.vm">
<parameters>
<parameter name="layout-name" value="default"/>
<parameter name="locale" value="en-UK"/>
<parameter name="controllerns" value="${controllerns}"/>
<parameter name="entityns" value="${entityns}"/>
<parameter name="generate-site-navigation" value="false"/>
<parameter name="permissions-group" value="partsbookeditors"/>
<parameter name="show-messages" value="true"/>
</parameters>
</style>
<split-regex in="${tmpdir}/views.auto.vm"
destdir="${views}" pattern="cut here: next file '([a-zA-Z0-9_./]*)'"/>
</target>