4:18 a.m.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml: rendered=#{!cons.refreshNeeded
and cons.active and cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and cons.active
and cons.producerInfo.registrationRequired is properly interpreted (or at the very least,
not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two expressions
!cons.refreshNeeded and cons.active), the EL engine still evaluates the last clause when
the state of the underlying objects is not ready for such an evaluation as the root
exception clearly states (java.lang.IllegalStateException: Registration status not yet
known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
4:22 a.m.
New subject: [JBoss JIRA] Resolved: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun resolved GTNWSRP-218.
----------------------------------
Resolution: Done
Thanks to Wesley's investigation, we have a workaround. However, I believe that the
underlying issue (the default EL evaluation) should be fixed at the AS/JSF (possibly PBR)
implementation level.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
7:04 a.m.
New subject: [JBoss JIRA] Reopened: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun reopened GTNWSRP-218:
----------------------------------
Turns out that the fix doesn't work on AS 5 so either we need a proper fix for AS 6 or
a different packaging of WSRP for AS 5 and AS 6.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
9:56 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Alexander Smirnov commented on GTNWSRP-218:
-------------------------------------------
To be honest, EL speck doesn't required boolean expressions to stop evaluation if
result can be calculated early ( first 'true' for or or 'false' for end),
unlike Java language spec. Therefore, you expectations about expression evaluation are
different with concrete implementation.
Anyway, I always suggest to avoid complicated EL expressions on JSF pages. They are slow,
hard to test and maintenance. If you need such calculations, create 'view model'
bean that represents visual state. Your page will be like :
rendered="#{viewBean.showRegistration} - simple, easy to understand.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
10:12 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun commented on GTNWSRP-218:
--------------------------------------
@Alexander: I believe your statement regarding the spec is wrong.
Per http://jsp.java.net/spec/jsp-2_1-fr-spec-el.pdf, section 1.9.1:
The operator stops as soon as the expression can be determined, i.e., A and B and C and D
– if B is false, then only A and B is evaluated.
Therefore, I maintain my statement that this is a bug in the EL evaluation engine. While I
agree that it could be better to introduce a bean method for this, I also think that the
expression should work as is.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
1:02 p.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Stan Silvert commented on GTNWSRP-218:
--------------------------------------
You should ask Remy about this. His team maintains the EL code.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
1:50 p.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Remy Maucherat commented on GTNWSRP-218:
----------------------------------------
I did not change anything related to the parsing or processing of "and". BTW, no
ready to use war test, no bug ;)
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
6:46 a.m.
New subject: [JBoss JIRA] Updated: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun updated GTNWSRP-218:
---------------------------------
Attachment: JSFRIPortlet.war
Test war that displays the issue.
Deploy in AS 6 and access: http://localhost:8080/JSFRIPortlet/jsf/home.jsf to replicate
the problem. Source available on demand.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
6:48 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun commented on GTNWSRP-218:
--------------------------------------
In essence, a simple JSF app with: <h:outputText rendered="#{!bean.a and
bean.b}">Foo</h:outputText>
bean is an instance of:
public class ELBug
{
public boolean isA()
{
return true;
}
public boolean isB()
{
throw new IllegalStateException("A is true so !A is false so B shouldn't be
evaluated in (!A and B)");
}
}
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
6:50 a.m.
New subject: [JBoss JIRA] Issue Comment Edited: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun edited comment on GTNWSRP-218 at 3/30/11 7:48 AM:
---------------------------------------------------------------
In essence, a simple JSF app with: <h:outputText rendered="#{!bean.a and
bean.b}">Foo</h:outputText>
bean is an instance of:
public class ELBug {
public boolean isA(){ return true; }
public boolean isB(){ throw new IllegalStateException("A is true so !A is false so
B shouldn't be evaluated in (!A and B)"); }
}
was (Author: claprun):
In essence, a simple JSF app with: <h:outputText rendered="#{!bean.a and
bean.b}">Foo</h:outputText>
bean is an instance of:
public class ELBug
{
public boolean isA()
{
return true;
}
public boolean isB()
{
throw new IllegalStateException("A is true so !A is false so B shouldn't be
evaluated in (!A and B)");
}
}
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
8:26 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Remy Maucherat commented on GTNWSRP-218:
----------------------------------------
Your example did not work for me, but I tested with AS 7 the following:
<jsp:useBean id="bean" scope="request"
class="org.gatein.test.ELBug" />
${!bean.a and bean.b}
It prints "false".
Really, looking at the evaluation code in the EL impl (which did not change from AS 5), I
don't understand how it would evaluate bean.b.
AstAnd reads:
public final class AstAnd extends BooleanNode {
public AstAnd(int id) {
super(id);
}
public Object getValue(EvaluationContext ctx)
throws ELException {
Object obj = children[0].getValue(ctx);
Boolean b = coerceToBoolean(obj);
if (!b.booleanValue()) {
return b;
}
obj = children[1].getValue(ctx);
b = coerceToBoolean(obj);
return b;
}
}
So if the left statement evaluates to false, it will return that without calling getValue
on the right portion of the statement. There may be relevant tests for that in the TCK as
well.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
8:47 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun commented on GTNWSRP-218:
--------------------------------------
What do you mean by: "did not work"? Were you not able to run the app or did you
it exhibit the issue?
I'm not interested in the behavior in AS 7 at the moment, it needs to run in AS 6.
Regarding the fact that the EL impl seems to be correct, I tend to agree with you since I
doubt that an issue as big as this one would have gone undetected until now. This might
have something to do with the PortletBridge… :/
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
8:47 a.m.
New subject: [JBoss JIRA] Assigned: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun reassigned GTNWSRP-218:
------------------------------------
Assignee: Wesley Hales (was: Chris Laprun)
Assigning to Wesley since it seems like the issue is originating in PortletBridge and not
the EL implementation bundled in AS 6.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Wesley Hales
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
9:22 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Remy Maucherat commented on GTNWSRP-218:
----------------------------------------
I did get some class not found related to portlets with the WAR (so I used a simpler JSP
instead). If I set the expression to ${bean.a and !bean.b}, I go get the following trace:
at org.gatein.test.ELBug.isB(ELBug.java:17) [classes:]
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) [:1.6.0_18]
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
[:1.6.0_18]
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
[:1.6.0_18]
at java.lang.reflect.Method.invoke(Method.java:616) [:1.6.0_18]
at javax.el.BeanELResolver.getValue(BeanELResolver.java:302)
[jboss-el-api_2.2_spec-1.0.0.Final.jar:1.0.0.Final]
at javax.el.CompositeELResolver.getValue(CompositeELResolver.java:175)
[jboss-el-api_2.2_spec-1.0.0.Final.jar:1.0.0.Final]
at org.apache.el.parser.AstValue.getValue(AstValue.java:134)
[jbossweb-7.0.0.Beta7.jar:7.0.0.Beta2-SNAPSHOT]
at org.apache.el.parser.AstNot.getValue(AstNot.java:42)
[jbossweb-7.0.0.Beta7.jar:7.0.0.Beta2-SNAPSHOT]
at org.apache.el.parser.AstAnd.getValue(AstAnd.java:42)
[jbossweb-7.0.0.Beta7.jar:7.0.0.Beta2-SNAPSHOT]
at org.apache.el.ValueExpressionImpl.getValue(ValueExpressionImpl.java:187)
[jbossweb-7.0.0.Beta7.jar:7.0.0.Beta2-SNAPSHOT]
at
org.apache.jasper.runtime.PageContextImpl.proprietaryEvaluate(PageContextImpl.java:935)
[jbossweb-7.0.0.Beta7.jar:7.0.0.Beta2-SNAPSHOT]
at org.apache.jsp.functions_jsp._jspService(functions_jsp.java:130)
So things are evaluating properly from what I can see. If the left portion of the and
evaluates to false, AstAnd should return without calling getValue on the right portion.
The stack trace you get should give clues I think.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Wesley Hales
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
9:28 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun commented on GTNWSRP-218:
--------------------------------------
at org.gatein.test.ELBug.isB(ELBug.java:17) [classes:] seems to imply that isB is
evaluated, which is definitely wrong. Am I missing something?
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Wesley Hales
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
9:42 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Remy Maucherat commented on GTNWSRP-218:
----------------------------------------
I did change the expression to show the stack trace.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Wesley Hales
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
10:06 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun commented on GTNWSRP-218:
--------------------------------------
Indeed, I see it now.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Wesley Hales
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
7:58 a.m.
New subject: [JBoss JIRA] Commented: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun commented on GTNWSRP-218:
--------------------------------------
Found the issue: GateIn pulls a dependency from shindig / gadget server:
addDependency(new Project("de.odysseus.juel", "juel-api",
"jar", "2.1.2")).
addDependency(new Project("de.odysseus.juel", "juel-impl",
"jar", "2.1.2"))
Apart from the fact that I'm not sure why the hell this is needed, this particular
version of JUEL isn't conform to the spec since, according to the release notes
(http://juel.sourceforge.net/project/changes.html):
developer: cbe context: code type: fix thanks to: Pavel Vojtechovsky. fixes: 2871795
Fixed: second operand is always evaluated in and/or operations (e.g. ${true or false}).
was fixed in 2.1.3/2.2.0-rc2 (released in October 2009), while we use 2.1.2 (released
2009/04/26!!!!). :(
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Wesley Hales
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
8:46 a.m.
New subject: [JBoss JIRA] Resolved: (GTNWSRP-218) Force use of JBoss EL on AS 6
[
https://issues.jboss.org/browse/GTNWSRP-218?page=com.atlassian.jira.plugi...
]
Chris Laprun resolved GTNWSRP-218.
----------------------------------
Assignee: Chris Laprun (was: Wesley Hales)
Resolution: Partially Completed
Needs to be fixed at the Portal level.
Force use of JBoss EL on AS 6
-----------------------------
Key: GTNWSRP-218
URL: https://issues.jboss.org/browse/GTNWSRP-218
Project: GateIn WSRP
Issue Type: Sub-task
Components: Admin GUI
Affects Versions: 2.1.0-Beta01
Reporter: Chris Laprun
Assignee: Chris Laprun
Fix For: 2.1.0-Beta02
Attachments: JSFRIPortlet.war
The EL evaluation engine seems to have changed between AS 5 and AS 6 and results in
improper evaluations of expressions leading to calls being made on objects that are not
ready to respond to them.
More specifically line 77 in /jsf/consumers/consumer.xhtml:
rendered=#{!cons.refreshNeeded and cons.active and
cons.producerInfo.registrationRequired}
Basically, it doesn't seem like the EL expression: !cons.refreshNeeded and
cons.active and cons.producerInfo.registrationRequired is properly interpreted (or at the
very least, not interpreted like it was on AS 5).
When the exception which causes the portlet to fail is thrown, the state is so that the
expression should not be completely evaluated:
cons.refreshNeeded is true so !cons.refreshNeeded is false, the evaluation of the
expression should stop because the result is already known to be false before looking at
the rest of the expression (which is the Java behavior, foo && bar results in bar
never being evaluated if foo is false).
Moreover, cons.active is also false (even though it should never be evaluated).
The exception is thrown because, despite the appropriate guards (the first two
expressions !cons.refreshNeeded and cons.active), the EL engine still evaluates the last
clause when the state of the underlying objects is not ready for such an evaluation as the
root exception clearly states (java.lang.IllegalStateException: Registration status not
yet known: call refresh first!). For me, this is a bug in the EL evaluation engine. At the
very least, this is a huge change in behavior which should be documented somewhere i.e.
the fact that you cannot rely on lazy evaluation of expressions in "and" clauses
but this seems to be too big a behavioral change (which would also go against how Java
works) for it to be proper behavior.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
4797
days inactive
4814
days old
18 comments
4 participants
participants (4)
-
Alexander Smirnov (JIRA) -
Chris Laprun (JIRA)
-
Remy Maucherat (JIRA)
-
Stan Silvert (JIRA)