Mail Archive Home | apollon-dev List | April 2003 Index
| Date Index | Thread Index |
Hi, I have tried to use Apollon (current CVS version) for the C-JDBC DTD. The processing fails and I get the following error: > ./apollon.sh ~/workspace/c-jdbc/c-jdbc/xml/c-jdbc.dtd Apollon: Apollon is processing the /home/mpeltier/workspace/c-jdbc/c-jdbc/xml/c-jdbc.dtd DTD... Please report the following exception trace to apollon-dev@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx System property os.name = Linux System property os.version = 2.4.18 System property java.version = 1.3.1_07 java.lang.IllegalArgumentException: An AtomicProperty cannot have a null XML type. at org.enhydra.zeus.binding.AtomicProperty.<init>(AtomicProperty.java:95) at org.enhydra.zeus.binding.AtomicProperty.<init>(AtomicProperty.java:166) at org.enhydra.zeus.binder.DTDBinder.handleContent(DTDBinder.java:354) at org.enhydra.zeus.binder.DTDBinder.handleContent(DTDBinder.java:391) at org.enhydra.zeus.binder.DTDBinder.handleContent(DTDBinder.java:305) at org.enhydra.zeus.binder.DTDBinder.convertToBinding(DTDBinder.java:217) at org.enhydra.zeus.binder.DTDBinder.getBindings(DTDBinder.java:153) at org.objectweb.apollon.ApollonProcessor.generate(ApollonProcessor.java:346) at org.objectweb.apollon.ApollonProcessor.process(ApollonProcessor.java:427) at org.objectweb.apollon.ApollonProcessor.process(ApollonProcessor.java:403) at org.objectweb.apollon.executor.ApollonApplication.process(ApollonApplication .java:318) at org.objectweb.apollon.executor.ApollonApplication.start(ApollonApplication.j ava:250) at org.objectweb.util.cmdline.lib.ApplicationBase.runMain(Unknown Source) at org.objectweb.apollon.executor.ApollonApplication.main(ApollonApplication.ja va:346) Could you help me to solve the pb? The DTD is parsed by C-JDBC (crimson) without problem. Thanks, Please find enclosed the DTD. -- Mathieu Peltier INRIA Rhône-Alpes - ObjectWeb Consortium http://www.inrialpes.fr/ - http://www.objectweb.org/
<?xml version="1.0" encoding="ISO-8859-1" ?>
<!--
/*
* C-JDBC: Clustered JDBC.
* Copyright (C) 2002, 2003 French National Institute For Research In Computer
* Science And Control (INRIA).
* Contact: c-jdbc@xxxxxxxxxxxxx
*
* This library is free software; you can redistribute it and/or modify it
* under the terms of the GNU Lesser General Public License as published by the
* Free Software Foundation; either version 2.1 of the License, or any later
* version.
*
* This library is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License
* for more details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with this library; if not, write to the Free Software Foundation,
* Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
*
* Initial developer(s): Emmanuel Cecchet.
* Contributor(s): ______________________________________.
*/
This is the XML DTD for C-JDBC controllers configuration files
Please use the following formal public identifier to identify it:
<!DOCTYPE C-JDBC PUBLIC "-//ObjectWeb//DTD C-JDBC x.y//EN"
"http://www.objectweb.org/c-jdbc/dtds/c-jdbc-x.y.dtd";>
Where x.y is the C-JDBC release number.
Or if you want to use a local copy of the DTD:
<!DOCTYPE C-JDBC SYSTEM "file:///path/to/c-jdbc-x.y.dtd">
For example:
Unix platforms:
<!DOCTYPE C-JDBC SYSTEM "file:///usr/local/c-jdbc/xml/c-jdbc-x.y.dtd">
Windows platforms:
<!DOCTYPE C-JDBC SYSTEM "file://c:/c-jdbc/xml/c-jdbc-x.y.dtd">
-->
<!-- ******************
Virtual Database
****************** -->
<!ELEMENT C-JDBC (VirtualDatabase+)>
<!--
A Virtual Database (vdb) is the database exposed to the user.
name: A vdb defined by a name that is used in the JDBC URL
(jdbc:cjdbc://host::/VirtualDatabaseName)
groupnName: If this VirtualDatabase is to be shared by several
C-JDBC controllers, a group name must be given so that the
controllers can synchronize and be aware of the requests received
by each of them.
portNumber: the port number on which clients (C-JDBC drivers) will
connect, default is 0 and means that the port number is chosen
automatically by the system.
backlogSize: the server socket backlog size (number of connections
that can wait in the accept queue before the system returns
'connection refused' to the client), default is 10.
maxNbOfConnections: maximum number of concurrent connections
accepted for this virtual database. Default is 0 (no limit)
poolThreads: if false, one thread is created for each connection
and dies when the connection closes. If set to true, threads are
reused to serve multiple connections. Default is true.
minNbOfThreads: minimum number of threads to keep in the pool (if
poolThreads is set to true), default is 0.
maxNbOfThreads: maximum number of threads in the pool (if
poolThreads is set to true), default is 0 (no limit).
maxThreadIdleTime: maximum time in seconds a thread can remain idle
before being removed from the pool. Default is 60s.
A VirtualDatabase contains :
- a set of real database backends,
- an Authentiation manager matches the virtual database and real
backends login/password,
- a request manager defines the behavior of the controller for this
VirtualDatabase.
The database schema can be enforced statically if dynamic discovery
does not work properly or is not desirable.
-->
<!ELEMENT VirtualDatabase (AuthenticationManager, DatabaseBackend+, RequestManager, DatabaseSchema?)>
<!ATTLIST VirtualDatabase
name CDATA #REQUIRED
groupName CDATA #IMPLIED
portNumber CDATA #IMPLIED
backlogSize CDATA #IMPLIED
maxNbOfConnections CDATA #IMPLIED
poolThreads (true | false) "true"
minNbOfThreads CDATA #IMPLIED
maxNbOfThreads CDATA #IMPLIED
maxThreadIdleTime CDATA #IMPLIED
>
<!-- ************************
Authentication manager
************************ -->
<!ELEMENT AuthenticationManager (AdminLogin, VirtualLogin+)>
<!--
An administrator login is used to access the management interface of a
virtual database and perform operations such as enabling/disabling
database backends, managing checkpoints, ...
-->
<!ELEMENT AdminLogin EMPTY>
<!ATTLIST AdminLogin
aLogin CDATA #REQUIRED
aPassword CDATA #REQUIRED
>
<!--
A virtual login is the login used by the client application through
the C-JDBC driver. The virtual login/password must match a "real"
login/password on each database backend.
-->
<!ELEMENT VirtualLogin (RealLogin+)>
<!ATTLIST VirtualLogin
vLogin CDATA #REQUIRED
vPassword CDATA #REQUIRED
>
<!--
Defines a login/password on a database backend identified by its
logical name.
-->
<!ELEMENT RealLogin EMPTY>
<!ATTLIST RealLogin
backendName CDATA #REQUIRED
rLogin CDATA #REQUIRED
rPassword CDATA #REQUIRED
>
<!-- ******************
Database Backend
****************** -->
<!--
A Database Backend is uniquely identified by its name/driver/url
name: a logical name identifying this backend
driver: the database native JDBC driver class name
url: the url to connect to give to the driver to connect to the
database
-->
<!ELEMENT DatabaseBackend (DatabaseSchema?, ConnectionManager+)>
<!ATTLIST DatabaseBackend
name CDATA #REQUIRED
driver CDATA #REQUIRED
url CDATA #REQUIRED
>
<!-- ****************************
Database schema definition
**************************** -->
<!--
A static DatabaseSchema can be defined to override the schema
automatically gathered by the controller. However, the schema must
remain compatible with the schema gathered from the backend.
-->
<!ELEMENT DatabaseSchema (DatabaseTable+)>
<!ELEMENT DatabaseTable (DatabaseColumn+)>
<!ATTLIST DatabaseTable
tableName CDATA #REQUIRED
nbOfColumns CDATA #REQUIRED
>
<!--
isUnique should be set to true if the column has a UNIQUE constraint.
This is the case for primary keys (composed primary keys are not yet
supported). This affects only cache behavior and select statements
parsing.
-->
<!ELEMENT DatabaseColumn EMPTY>
<!ATTLIST DatabaseColumn
columnName CDATA #REQUIRED
isUnique (true | false) "false"
>
<!-- *********************
Connection Managers
********************* -->
<!--
One Connection Manager must be defined for each virtual login the
backend belongs to. The user/password combination defined in the
RealLogin element for this vLogin/Backend is used to create the
Connection Manager.
-->
<!ELEMENT ConnectionManager (SimpleConnectionManager | FailFastPoolConnectionManager | RandomWaitPoolConnectionManager | VariablePoolConnectionManager)>
<!ATTLIST ConnectionManager
vLogin CDATA #REQUIRED
>
<!--
SimpleConnectionManager is a basic Connection Manager that opens a
new connection on each request and closes it at the end. It could be
useful if the underlying driver already implements connection pooling
for example.
-->
<!ELEMENT SimpleConnectionManager EMPTY>
<!--
FailFastPoolConnectionManager offers connection pooling and fail fast
when the pool is empty.
poolSize is the size of the pool.
All connections are initialized at startup time and if the pool size
is too large it is adjusted to the largest number of connections
available.
Once the pool is empty, null is returned instead of a
connection. Therefore incoming requests will fail until at least one
connection is freed. No system overload should occur with this
connection manager, but if the pool size is too small, many requests
will fail.
-->
<!ELEMENT FailFastPoolConnectionManager EMPTY>
<!ATTLIST FailFastPoolConnectionManager
poolSize CDATA #REQUIRED
>
<!--
RandomWaitPoolConnectionManager provides connection pooling and wait
when the pool is empty.
poolSize is the size of the pool.
timeout is the maximum time to wait for a connection in seconds (0
is default and means no timeout: wait until one connection is
freed).
All connections are initialized at startup time and if the pool size
is too large it is adjusted to the largest number of connections
available.
Once the pool is empty, the requests wait until a connection is freed
or the specified timeout has elapsed. The FIFO order of connection
request is not ensured by this connection manager.
-->
<!ELEMENT RandomWaitPoolConnectionManager EMPTY>
<!ATTLIST RandomWaitPoolConnectionManager
poolSize CDATA #REQUIRED
timeout CDATA #IMPLIED
>
<!--
VariablePoolConnectionManager provides connection pooling with a
dynamically adjustable pool size.
initPoolSize is the initial pool size to be intialized at startup.
minPoolSize is the minimum pool size (default is equal to the
initPoolSize).
maxPoolSize is the maximum pool size (0 is the default and means no
limit).
idleTimeout is the time in seconds a connection can stay idle
before being released (removed from the pool). Default is no
timeout which means that once allocated, connections are never
released.
waitTimeout is the maximum time in seconds to wait for a connection
in seconds (0 is default and means no timeout=wait until one
connection is freed).
-->
<!ELEMENT VariablePoolConnectionManager EMPTY>
<!ATTLIST VariablePoolConnectionManager
initPoolSize CDATA #REQUIRED
minPoolSize CDATA #IMPLIED
maxPoolSize CDATA #IMPLIED
idleTimeout CDATA #IMPLIED
waitTimeout CDATA #IMPLIED
>
<!-- *****************
Request Manager
***************** -->
<!--
The Request Manager is composed of a scheduler, an optional cache, a
load balancer and an optional recovery log.
If requests need to be parsed, it can be done sequentially when
needed (backgroundParsing is set to false which is the default value)
or forced to be performed in background by a separate thread (it
means a new thread is created for each request that need to be
parsed). A timeout in seconds can be defined for
begin/commit/rollback operations. If no value is given, the default
timeout is set to 60 seconds. Warning! 0 means no timeout and wait
forever until completion.
-->
<!ELEMENT RequestManager (RequestScheduler, RequestCache?, LoadBalancer, RecoveryLog?)>
<!ATTLIST RequestManager
backgroundParsing (true | false) #IMPLIED
beginTimeout CDATA #IMPLIED
commitTimeout CDATA #IMPLIED
rollbackTimeout CDATA #IMPLIED
>
<!-- ********************
Request Schedulers
******************** -->
<!ELEMENT RequestScheduler (SingleDBScheduler | RAIDb-0Scheduler | RAIDb-1Scheduler | RAIDb-2Scheduler)>
<!ELEMENT SingleDBScheduler EMPTY>
<!ATTLIST SingleDBScheduler
level (query | optimisticTransaction | pessimisticTransaction) #REQUIRED
>
<!ELEMENT RAIDb-0Scheduler EMPTY>
<!ATTLIST RAIDb-0Scheduler
level (query | pessimisticTransaction) #REQUIRED
>
<!ELEMENT RAIDb-1Scheduler EMPTY>
<!ATTLIST RAIDb-1Scheduler
level (query | optimisticQuery | optimisticTransaction | pessimisticTransaction) #REQUIRED
>
<!ELEMENT RAIDb-2Scheduler EMPTY>
<!ATTLIST RAIDb-2Scheduler
level (query | pessimisticTransaction) #REQUIRED
>
<!-- ****************
Request Caches
**************** -->
<!ELEMENT RequestCache (QueryCache | QueryCacheDB)>
<!--
The query cache provides both query result caching and query parsing
caching. If two exact same requests are to be executed, only one is
executed and the second one waits until the completion of the first
one (this is the default pendingTimeout value which is 0). To prevent
the second request to wait forever, a pendingTimeout value in seconds
can be defined for the waiting request. If the timeout expires the
request is executed in parallel of the first one.
noInvalidation provides a non-coherent cache and should only be used
for testing purposes. table and column provide table-based and
column-based invalidations respectively. columnUnique can optimize
requests that select a unique primary key.
-->
<!ELEMENT QueryCache EMPTY>
<!ATTLIST QueryCache
pendingTimeout CDATA #IMPLIED
granularity (noInvalidation | table | column | columnUnique) #REQUIRED
>
<!--
The following query caches is experimental and use a separate
database for advanced query caching processing.
-->
<!ELEMENT QueryCachetDB EMPTY>
<!ATTLIST QueryCacheDB
driver CDATA #REQUIRED
url CDATA #REQUIRED
login CDATA #REQUIRED
password CDATA #REQUIRED
granularity (columnUniqueInsertDB|columnUniqueInsertDBUniqueUpdate) "columnUniqueInsertDB"
>
<!-- ****************
Load Balancers
**************** -->
<!ELEMENT LoadBalancer (SingleDB | RAIDb-0 | RAIDb-1 | RAIDb-1ec | RAIDb-2 | RAIDb-2ec)>
<!--
No RAIDb: Just a single database backend
-->
<!ELEMENT SingleDB EMPTY>
<!--
RAIDb-0: full partitioning
If CreateTable is omitted, new table creation will use a round-robin
algorithm on all backends.
-->
<!ELEMENT RAIDb-0 (CreateTable*)>
<!--
RAIDb-1: full mirroring
If WaitForCompletion is omitted, the default behaviour is to return
the result as soon as one backend has completed.
-->
<!ELEMENT RAIDb-1 (WaitForCompletion?, (RAIDb-1-RoundRobin | RAIDb-1-WeightedRoundRobin | RAIDb-1-LeastPendingRequestsFirst))>
<!ELEMENT RAIDb-1-RoundRobin EMPTY>
<!ELEMENT RAIDb-1-WeightedRoundRobin (BackendWeight)>
<!ELEMENT RAIDb-1-LeastPendingRequestsFirst EMPTY>
<!--
RAIDb-1ec: full mirroring with error checking
Error checking here means that read requests will be sent to several
nodes and results will be compared to detect byzantine failures (see
Error cheking policy below).
If WaitForCompletion is omitted, the default behaviour is to return
the result as soon as one backend has completed (on writes).
nbOfConcurrentReads represents the number of read than can occur in
parallel.
Note that nbOfConcurrentReads threads will be created per backend.
-->
<!ELEMENT RAIDb-1ec (WaitForCompletion?, ErrorChecking, (RAIDb-1ec-RoundRobin | RAIDb-1ec-WeightedRoundRobin))>
<!ATTLIST RAIDb-1ec
nbOfConcurrentReads CDATA #REQUIRED
>
<!ELEMENT RAIDb-1ec-RoundRobin EMPTY>
<!ELEMENT RAIDb-1ec-WeightedRoundRobin (BackendWeight)>
<!--
RAIDb-2: distributed mirroring
If CreateTable is omitted, new table creation will affect all backends.
If WaitForCompletion is omitted, the default behaviour is to return
the result as soon as one backend has completed.
-->
<!ELEMENT RAIDb-2 (CreateTable*, WaitForCompletion?, (RAIDb-2-RoundRobin | RAIDb-2-WeightedRoundRobin | RAIDb-2-LeastPendingRequestsFirst))>
<!ELEMENT RAIDb-2-RoundRobin EMPTY>
<!ELEMENT RAIDb-2-WeightedRoundRobin (BackendWeight)>
<!ELEMENT RAIDb-2-LeastPendingRequestsFirst EMPTY>
<!--
RAIDb-2ec: distributed mirroring with error checking
Error checking here means that read requests will be sent to several
nodes and results will be compared to detect byzantine failures (see
Error cheking policy below).
If WaitForCompletion is omitted, the default behaviour is to return
the result as soon as one backend has completed (on writes).
nbOfConcurrentReads represents the number of read than can occur in
parallel.
Note that nbOfConcurrentReads threads will be created per backend.
-->
<!ELEMENT RAIDb-2ec (CreateTable*, WaitForCompletion?, ErrorChecking, (RAIDb-2ec-RoundRobin | RAIDb-2ec-WeightedRoundRobin))>
<!ATTLIST RAIDb-2ec
nbOfConcurrentReads CDATA #REQUIRED
>
<!ELEMENT RAIDb-2ec-RoundRobin EMPTY>
<!ELEMENT RAIDb-2ec-WeightedRoundRobin (BackendWeight)>
<!--
Policy to adopt when waiting for the completion of a request. Policy
works as follows:
- first: return the result as soon as one node has completed.
- majority: return the result as soon as a majority of nodes
(n/2+1) has completed.
- all: wait for all nodes to complete.
-->
<!ELEMENT WaitForCompletion EMPTY>
<!ATTLIST WaitForCompletion
policy (first | majority | all) "first"
>
<!--
Error checking policy (for RAIDb-1ec and RAIDb2-ec). Error checking
is used to detect byzantine failures of nodes. It means detecting
when a node sends funny results in a non- deterministic way. Error
checking allows read queries to be sent to more than one database,
and the results are compared. A majority of nodes must agree on the
result that will be sent to the client. Error checking policies are
defined as follows:
- random: numberOfNodes backends are picked up randomly; the read
request is sent to these backends and results are compared.
- roundRobin: numberOfNodes backends are picked up using a
Round-Robin algorithm ; the read request is send to these backends
and results are compared.
- all: the request is sent to ALL nodes (numberOfNodes is ignored)
and the results compared.
numberOfNodes must be greater or equal to 3.
-->
<!ELEMENT ErrorChecking EMPTY>
<!ATTLIST ErrorChecking
policy (random | roundRobin | all) #REQUIRED
numberOfNodes CDATA #REQUIRED
>
<!--
Policy to adopt when creating a new table. This policy is based on
the given list of BackendName nodes (which might be a subset of the
complete set of backends). If the backend list is omitted, then all
enabled backends are taken at decision time. numberOfNodes represent
the number of backends to pickup from the BackendName list to apply
the policy (it must be set to 1 for RAIDb-0 load balancers and can
never be greater than the number of nodes declared in the BackendName
list). If tableName is omitted, it is considered as a default policy
to be used for all create table statements that don't have a specific
rule. Else tableName specifies a name pattern to which this policy
applies.
Policy works as follows:
- random: numberOfNodes backends are picked up randomly from the
BackendName list and the table is created on these nodes.
- roundRobin: numberOfNodes backends are picked up from the
BackendName list using a Round-Robin algorithm and the table is
created on these nodes.
- all: the table is created on ALL nodes in the BackendName list
(numberOfNodes is ignored).
-->
<!ELEMENT CreateTable (BackendName*)>
<!ATTLIST CreateTable
tableName CDATA #IMPLIED
policy (random | roundRobin | all) #REQUIRED
numberOfNodes CDATA #REQUIRED
>
<!--
BackendName simply identifies a backend by its logical name
-->
<!ELEMENT BackendName EMPTY>
<!ATTLIST BackendName
name CDATA #REQUIRED
>
<!--
BackendWeight defines the weight of a backend for weighted
round-robin algorithms
name: the logical name of the backend
weight: the weight expressed as an integer
-->
<!ELEMENT BackendWeight EMPTY>
<!ATTLIST BackendWeight
name CDATA #REQUIRED
weight CDATA #REQUIRED
>
<!-- **************
Recovery Log
************** -->
<!ELEMENT RecoveryLog (FileRecoveryLog|JDBCRecoveryLog)>
<!ELEMENT FileRecoveryLog EMPTY>
<!ATTLIST FileRecoveryLog
fileName CDATA #REQUIRED
>
<!--
A timeout in seconds can be defined for the sql requests. If no
value is given, the default timeout is set to 60 seconds. Warning! 0
means no timeout and wait forever until completion.
-->
<!ELEMENT JDBCRecoveryLog EMPTY>
<!ATTLIST JDBCRecoveryLog
driver CDATA #REQUIRED
url CDATA #REQUIRED
login CDATA #REQUIRED
password CDATA #REQUIRED
logTable CDATA #REQUIRED
checkpointTable CDATA #REQUIRED
requestTimeout CDATA #IMPLIED
>
| Date Index | Thread Index |
Powered by MHonArc.
Copyright © 1999-2005, ObjectWeb Consortium | contact | webmaster.