March 2008 - jboss-dev-forums - Jboss List Archives

[Design of JBoss Transaction Services] - Re: TransactionTimeoutConfiguration pluggability

by adrian＠jboss.org

"jhalliday" wrote : The tin actually contains no specification to cover this - it only requires that the exception is thrown if the tx is marked for rollback, not if it's actually rolled back or rolling back. Basically the interface spec does not anticipate async rollback processing. I could close the JIRA as 'this is not a bug' until the spi javadoc is revised. The javadoc was written against the old TM behaviour it obviosuly needs updating. RolledBack/Committed/Completed/etc. is also an error (only RollbackOnly was possible with the old TM). anonymous wrote : Or in light of your comment here, I could throw away the JBossTS implementation of TransactionTimeoutConfiguration and nothing in the AS would break? Correct, except for people that expect <set-tx-timeout/> in the jdbc resource adapter to work like it used to. Instead no query timeout will be set automatically (not fatal, but it is a regression of sorts). P.S. I've got a "feature request" open for JavaEE6 that this method or something like it gets added to the TransactionSynchronizationRegistry (JTA spec). * It will mean this behaviour is portable across appservers * You'd have to implement it anyway ;-) View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137799#4137799 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137799

18 years

1
0
0 / 0

[Design the new POJO MicroContainer] - Re: JarEntry as VFS root does not return empty vfspath

by scott.stark＠jboss.org

Currently the name is "" for the root is the contract. Make the difference between the name and vfs path context name explicit to document the different, JBVFS-19 | /** | * Get the simple VF name (X.java) | * | * @return the simple, non-empty file name of the VF | */ | String getName(); | | /** | * Get the simple VF context name (X.java). This is the name portion of the VF in the pathName. | * | * @return the path context name for the VFS relative path name. This will be empty for root VFs. | */ | String getContextName(); | | /** | * Get the VFS relative path name (org/jboss/X.java) | * | * @return the VFS relative path name. This will be empty for root VFs. | */ | String getPathName(); | View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137797#4137797 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137797

18 years

1
0
0 / 0

[Design of JBoss Transaction Services] - Re: TransactionTimeoutConfiguration pluggability

by jhalliday

The tin actually contains no specification to cover this - it only requires that the exception is thrown if the tx is marked for rollback, not if it's actually rolled back or rolling back. Basically the interface spec does not anticipate async rollback processing. I could close the JIRA as 'this is not a bug' until the spi javadoc is revised. Or in light of your comment here, I could throw away the JBossTS implementation of TransactionTimeoutConfiguration and nothing in the AS would break? View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137794#4137794 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137794

18 years

1
0
0 / 0

[Design of JBoss Transaction Services] - Re: TransactionTimeoutConfiguration pluggability

by adrian＠jboss.org

"jhalliday" wrote : > It is an optional extension interface to the TransactionManager. | | In that case should it not be declared as | | public interface TransactionTimeoutConfiguration extends TransactionManager | No. Because the ConnectionManager only implements TTC not the full TM. anonymous wrote : | Right now nothing in the integration spi that forces implementations to use the the same object to provide both. Therefore making that assumption in the test is a bad thing. | If it doesn't implement the interface then don't run the test. | public void testSomething() | { | Object tm = TransactionManagerLocate.locate(); | if (tm instanceof TransactionTimeoutConfiguration) | return; // The interface is optional | } | But if you do implement the interface then it should "do what it says on the tin" ;-) View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137791#4137791 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137791

18 years

1
0
0 / 0

[Design the new POJO MicroContainer] - AbstractTestCase::assterInstanceOf is broken

by alesj

Dunno how this slipped by us for so long: | protected <T> T assertInstanceOf(Object o, Class<T> expectedType, boolean allowNull) | { | if (expectedType == null) | fail("Null expectedType"); | | if (o == null && allowNull) | return null; | | try | { | return expectedType.cast(o); | } | catch (ClassCastException e) | { | fail("Object " + o + " of class " + o.getClass().getName() + " is not an instanceof " + expectedType.getName()); | // should not reach this | return null; | } | } | What about if allowNull == false and o == null? :-( A test for test required. :-) View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137790#4137790 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137790

18 years

1
0
0 / 0

[Design of JBoss Build System] - Re: Shared Parent POM for dependencyManagement

by pgier

"pgier" wrote : "ALRubinger" wrote : I have committed org.jboss.jbossas:jboss-as-dependency: | | | | http://jira.jboss.com/jira/browse/JBAS-5324 | | | | Looking to avoid being too invasive here. The shared POM is now committed and deployed for review, though I'm holding the commit for org.jboss.jbossas:jboss-as-parent until we can verify that all looks allright. Once Paul gives me the nod, I'll commit that change. | | | | The patch up for review is attached to the JIRA, as is the commit of the new project. | | | | Paul, let's look over this together. | | | | S, | | ALR | | Hey Andrew, I think it's fine to commit this, except for the prerequisites section (maven 2.0.8). I think that should stay in the jbossas parent because it doesn't get inherited. If ejb also requires 2.0.8 you'll have to define it in that parent also. | | Also, I must have committed the 2.2-beta-3-SNAPSHOT version of the assembly plugin by mistake. I was working with that locally, but the one in the pom should be 2.2-beta-2. | | One more thing I realized, is that the jbossas intermodule dependencies (deps with groupid "org.jboss.jbossas") will probably have to remain in the as parent, and not in the new super parent. Otherwise we could have a chicken/egg problem when trying to release. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137787#4137787 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137787

18 years

1
0
0 / 0

[Design of JBoss Transaction Services] - Re: TransactionTimeoutConfiguration pluggability

by jhalliday

> It is an optional extension interface to the TransactionManager. In that case should it not be declared as public interface TransactionTimeoutConfiguration extends TransactionManager Right now nothing in the integration spi that forces implementations to use the the same object to provide both. Therefore making that assumption in the test is a bad thing. > Either leave it failing with a "FIXME JBAS-5081" in the error message that the test produces. > Or if it is stopping a release add it to the "bad.tests.exclude" section. Well it is going into AS branch 4.2 initially and since we don't have any releases planned from that, leaving it failing should not be a problem. Trunk is more of a problem, I don't know if our new JBossTS release will be out before AS 5 CR1. Probably best to exclude it for that. Thanks View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137786#4137786 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137786

18 years

1
0
0 / 0

[Design of JBoss Build System] - Re: Shared Parent POM for dependencyManagement

by adrian＠jboss.org

"ALRubinger" wrote : | So we can also look at locking down versions by placing them in brackets, forcing a version conflict failure if there's a problem. Brian alludes to some problems here; I'm not sure in which way these might manifest themselves. | No we should use constraints (version ranges). But we need to understand how that interacts with the release process and reproducability of tagged versions. Specific versions is just a very restrictive form of version constraint that will make it virtually impossible to upgrade projects without also doing dummy releases of other projects to change their constraints to agree. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137782#4137782 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137782

18 years

1
0
0 / 0

[Design of JBoss Build System] - Re: Shared Parent POM for dependencyManagement

by pgier

"pgier" wrote : | | Also, I must have committed the 2.2-beta-3-SNAPSHOT version of the assembly plugin by mistake. I was working with that locally, but the one in the pom should be 2.2-beta-2. | | Sorry, you can ignore this part, since I see it is already set to 2.2-beta-2. I misunderstood some messages in my local environment. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137781#4137781 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137781

18 years

1
0
0 / 0

[Design of JBoss Transaction Services] - Re: TransactionTimeoutConfiguration pluggability

by adrian＠jboss.org

"jhalliday" wrote : Adrian and the JBossAS team: | | I'm fixing http://jira.jboss.com/jira/browse/JBAS-5081 for use in JBossAS 4.2 and will then port across to 5.0 | | I'm trying to get my head around how to write JBossAS testsuite tests, specifically one that will cover this. | | I don't see any way in the AS to get the object that implements TransactionTimeoutConfiguration in a pluggable fashion. The best I can do is assume that the TransactionManager also implements it and cast appropriately. Do we need a TransactionTimeoutConfigurationLocator to live alongside of the TransactionManagerLocator? | It is an optional extension interface to the TransactionManager. It also gets implemented by the JBossJCA TxConnectionManager by forwarding it to the tm (if it implements it). anonymous wrote : | Also, what do I do in the case of a new test like this that I know is going to fail until the next JBossTS release is available for consumption by the AS? Check it in anyhow and accept that the testsuite won't pass for a few weeks? | | Thanks Either leave it failing with a "FIXME JBAS-5081" in the error message that the test produces. Or if it is stopping a release add it to the "bad.tests.exclude" section. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4137777#4137777 Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4137777

18 years

1
0
0 / 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

jboss-dev-forums March 2008