Re: jbpm-3.2.5.SP
by Thomas Diesler
what jbpm column is that?
It should be easy enough to reproduce this in a dedicated test.
Julian Coleman wrote:
> Hi,
>
>> Thanks. Do we have a test case that actually uses long strings?
>
> Just to confirm, if I modify the SOA bpm_orchestration1 sample to have a
> 270 character message, I see:
>
> org.h2.jdbc.JdbcSQLException: Value too long for column STRINGVALUE_
>
> exceptions, so I will be able to verify a fix. Unfortunately, the "easy
> way for projects to build/test the SOA platform" is still a few weeks
> away (at least).
>
> Thanks,
>
> J
>
16 years, 8 months
Re: jbpm-3.2.5.SP
by Thomas Diesler
Hi Julian,
are you actually changing stuff at that level?
We verify jBPM functionality and database backward compatibility with
the hibernate mapping that is in place for a given version. We also
generate SQL update scripts from those if needed.
A change at that level is not cosmetic and effectively invalidates our QA.
Are you saying that this issue needs to be reopened?
StringInstance hbm configuration column length of 4000
https://jira.jboss.org/jira/browse/JBPM-1109
Here is the changeset that removed length="4000"
http://fisheye.jboss.org/changelog/JbpmSvn?cs=2440
Maybe Alejandro can fill in the details why length="4000" was removed.
-thomas
Julian Coleman wrote:
> Hi,
>
>> After you've taken the stuff back out again, could you please start
>> another QA build for 3.2.5.SP? We can can only tag from a clean QA build.
>
> There is another blocking JIRA for SOA 4.3.0 CP01/jBPM 3.2.5 SP1 - SOA-1164.
> This is a new incarnation of the bug orginally discussed in SOA-271,
> JBESB-1435 and JBPM-1109.
>
> I am testing a build with the fix from JBPM-1109 applied to:
>
> modules/core/src/main/resources/org/jbpm/context/exe/variableinstance/StringInstance.hbm.xml
>
> but I am unsure if:
>
> modules/core/src/main/resources/org/jbpm/context/exe/variableinstance/HibernateStringInstance.hbm.xml
> modules/core/src/main/resources/org/jbpm/context/log/variableinstance/StringUpdateLog.hbm.xml
>
> also need to be changed (or, indeed, if I have missed another location). Any
> help/confirmation appreciated. We can't tag SP1 until this is fixed.
>
> Thanks,
>
> J
>
16 years, 8 months
Re: jbpm-3.2.5.SP
by Thomas Diesler
For your reference: https://jira.jboss.org/jira/browse/JBPM-2018
Could you please comment on the issue with the exact requirements that
should actually be checked for. i.e. which jbpm fields does this issue
apply to?
-thomas
Julian Coleman wrote:
> Hi,
>
>> are you actually changing stuff at that level?
>
> We don't make any changes here in the SOA platform build
>
>> A change at that level is not cosmetic and effectively invalidates our QA.
>
> Indeed. So this change will require a re-test.
>
>> Are you saying that this issue needs to be reopened?
>
> Yes. The previous SOA version had a length of 4000, so this one needs it as
> well.
>
>> Here is the changeset that removed length="4000"
>> http://fisheye.jboss.org/changelog/JbpmSvn?cs=2440
>
> Thanks for the pointer. I've tried reversing the removal of the length 4000
> parts of that changeset, but the generated sql still has (e.g.) varchar(255)
> and nothing with length 4000, so I must be missing something.
>
>> Maybe Alejandro can fill in the details why length="4000" was removed.
>
> That might be useful. Somehow, we need it back though ;-)
>
> Thanks,
>
> J
>
16 years, 8 months
Re: jbpm-3.2.5.SP
by Thomas Diesler
You mean jbpm-3.2.3 and jpdl-3.2.2-SOA-4.2.0.CP03 ?
svn diff https://svn.jboss.org/repos/jbpm/jbpm3/tags/jpdl-3.2.3.GA
https://svn.jboss.org/repos/jbpm/jbpm3/tags/jpdl-3.2.2-SOA-4.2.0.CP03 >
jpdl-3.2.3.GA_jpdl-3.2.2-SOA-4.2.0.CP03.diff
.. so far, this command has not come back on my box
-thomas
Mark Little wrote:
> Thomas, can you do an svn diff between the baseline of your tree and the
> version of jBPM we have been shipping in the platform?
>
> Mark.
>
>
> On 5 Feb 2009, at 12:32, Thomas Diesler wrote:
>
>>> We all assumed that was the case, but this issue raises
>>> uncertainties. There is nothing that has higher priority than doing
>>> this today as it is delaying the release, and if other bug fixes are
>>> missing we need a schedule from you on when we can get a new tree
>>> that incorporates them all.
>>
>> Yes, lets first find out what actually needs doing.
>
> ---
> Mark Little
> mlittle(a)redhat.com
>
> JBoss, a Division of Red Hat
> Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod
> Street, Windsor, Berkshire, SI4 1TE, United Kingdom.
> Registered in UK and Wales under Company Registration No. 3798903
> Directors: Michael Cunningham (USA), Charlie Peters (USA), Matt Parsons
> (USA) and Brendan Lane (Ireland).
>
>
>
16 years, 8 months
Re: jbpm-3.2.5.SP
by Thomas Diesler
The build should ok. What you are seeing is an attempt to generates SQL
update scripts based on an existing 3.2.2 schema. I suppose you don't
have a 3.2.2 sybase schema that you can point to when doing the build.
Julian Coleman wrote:
> Hi,
>
>> I am testing a build with the fix from JBPM-1109 applied to:
>
> Looking through the jBPM build and the logs in the area related to the DB
> schemas, I see:
>
> [exec] update-schema:
> ...
> [exec] [jbpmschema] java.sql.SQLException: Illegal connection port value '${jdbc.mysql.port}'
> ...
> [exec] [jbpmschema] java.sql.SQLException: The syntax of the connection URL 'jdbc:jtds:sybase://${jdbc.sybase.server}:${jdbc.sybase.port}/jbpm322' is invalid.
>
> so something is going wrong here too. Unfortunately, it doesn't stop the
> build (I didn't notice it until I looked through the build log).
>
> Thanks,
>
> J
>
16 years, 8 months
Re: jbpm-3.2.5.SP
by Thomas Diesler
Mark Little wrote:
> Thanks Trev and J for looking into this.
>
> Thomas, we need to be sure that *all* of the issues that we had fixed in
> the previous jBPM version we have been shipping in the SOA-P are fixed
> in the version you are providing.
My baseline for mavenization was jbpm-3.2.3. could you please talk to
Koen to verify whether the above assumption is correct?
> We all assumed that was the case, but
> this issue raises uncertainties. There is nothing that has higher
> priority than doing this today as it is delaying the release, and if
> other bug fixes are missing we need a schedule from you on when we can
> get a new tree that incorporates them all.
Yes, lets first find out what actually needs doing.
>
> Thanks,
>
> Mark.
>
>
> On 5 Feb 2009, at 11:20, trevor kirby wrote:
>
>> Thomas, According to JBPM-1109 this fix was committed to both the SOA
>> branch and HEAD so it should have made it into 3.2.4 et al. As a
>> matter of some urgency can you please check that no other bug fixes
>> have been missed in the mavenisation process and that there are no
>> other incompatibilities introduced.
>>
>> Trev
>>
>> Julian Coleman wrote:
>>
>>
>>> Hi,
>>>
>>>
>>>> After you've taken the stuff back out again, could you please start
>>>> another QA build for 3.2.5.SP? We can can only tag from a clean QA
>>>> build.
>>>>
>>>
>>> There is another blocking JIRA for SOA 4.3.0 CP01/jBPM 3.2.5 SP1 -
>>> SOA-1164.
>>> This is a new incarnation of the bug orginally discussed in SOA-271,
>>> JBESB-1435 and JBPM-1109.
>>>
>>> I am testing a build with the fix from JBPM-1109 applied to:
>>>
>>> modules/core/src/main/resources/org/jbpm/context/exe/variableinstance/StringInstance.hbm.xml
>>>
>>>
>>> but I am unsure if:
>>>
>>> modules/core/src/main/resources/org/jbpm/context/exe/variableinstance/HibernateStringInstance.hbm.xml
>>>
>>> modules/core/src/main/resources/org/jbpm/context/log/variableinstance/StringUpdateLog.hbm.xml
>>>
>>>
>>> also need to be changed (or, indeed, if I have missed another
>>> location). Any
>>> help/confirmation appreciated. We can't tag SP1 until this is fixed.
>>>
>>> Thanks,
>>>
>>> J
>>>
>>>
>>
>
> ---
> Mark Little
> mlittle(a)redhat.com
>
> JBoss, a Division of Red Hat
> Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod
> Street, Windsor, Berkshire, SI4 1TE, United Kingdom.
> Registered in UK and Wales under Company Registration No. 3798903
> Directors: Michael Cunningham (USA), Charlie Peters (USA), Matt Parsons
> (USA) and Brendan Lane (Ireland).
>
>
>
16 years, 8 months
Re: jbpm-3.2.5.SP
by Alejandro Guizar
Update to my previous email: I ran the test matrix against 3.2.5 SP with
version columns commented out. Several tests having some form of
concurrency failed.
http://jbpm.dyndns.org:8180/hudson/job/jBPM3-Database-Matrix/1/
Without version checking, a job may get executed more than once. Hence
pure pessimistic locking is not possible, at least not the way jobs are
acquired currently.
Please go ahead and tag 3.2.5 SP1. I will take JBPM-1953 out of that
release and discuss the alternatives with Thomas.
-Alejandro
El mié, 04-02-2009 a las 14:48 +0000, Julian Coleman escribió:
> Hi,
>
> Thanks for looking at these.
>
> > Today I ported JBPM-1903 to 3.2.5.SP because it was relevant to the
> > upcoming migration to JMS.
> >
> > Regarding JBPM-1452, the enterprise services have to be enabled in the
> > SOA-P jBPM configuration file. I made a few remarks to Jiri's setup
> > instructions directly in JIRA.
>
> For this, we just need to update our JBPM Reference Guide to take into
> account these changes.
>
> > As for JBPM-1453, the feature requested there is unfeasible. We cannot
> > disable optimistic locking with a switch. We would need an alternate JAR
> > with no version columns declared in the mapping files.
>
> So, we won't get a fix for this for SOA 4.3.0 CP01. Are there any other
> issues that need to be pulled up to the JBPM 3.2.5 SP branch? If not, can
> we tag 3.2.5 SP1 and build the next SOA release candidate from it?
>
> Thanks,
>
> J
>
16 years, 8 months
Re: jbpm-3.2.5.SP
by Alejandro Guizar
Hi Julian, please read my comments below.
El mié, 04-02-2009 a las 14:48 +0000, Julian Coleman escribió:
> Hi,
>
> Thanks for looking at these.
>
> > Today I ported JBPM-1903 to 3.2.5.SP because it was relevant to the
> > upcoming migration to JMS.
> >
> > Regarding JBPM-1452, the enterprise services have to be enabled in the
> > SOA-P jBPM configuration file. I made a few remarks to Jiri's setup
> > instructions directly in JIRA.
>
> For this, we just need to update our JBPM Reference Guide to take into
> account these changes.
Since 3.2.4 the jBPM user guide contains a new chapter on application
server facilities. I also removed the old content on the subject. Let me
know if I can help further in the documentation area.
> > As for JBPM-1453, the feature requested there is unfeasible. We cannot
> > disable optimistic locking with a switch. We would need an alternate JAR
> > with no version columns declared in the mapping files.
>
> So, we won't get a fix for this for SOA 4.3.0 CP01. Are there any other
> issues that need to be pulled up to the JBPM 3.2.5 SP branch? If not, can
> we tag 3.2.5 SP1 and build the next SOA release candidate from it?
Before you do, I want to run a test matrix against the 3.2.5 SP branch
without version columns. If it passes, then we can confidently provide
our customers a jBPM build that will not produce optimistic locking
exceptions.
> Thanks,
>
> J
>
16 years, 8 months
Re: jbpm-3.2.5.SP
by Alejandro Guizar
Oh... no, I meant JBPM-1952 and JBPM-1953 respectively. Thanks for the
observation.
-Alejandro
El mié, 04-02-2009 a las 10:49 +0000, Mark Little escribió:
> Hi Alejandro. Did you really mean JBPM-1453?
>
> Mark.
>
>
> On 4 Feb 2009, at 06:48, Alejandro Guizar wrote:
>
> > Today I ported JBPM-1903 to 3.2.5.SP because it was relevant to the
> > upcoming migration to JMS.
> >
> > Regarding JBPM-1452, the enterprise services have to be enabled in the
> > SOA-P jBPM configuration file. I made a few remarks to Jiri's setup
> > instructions directly in JIRA.
> >
> > As for JBPM-1453, the feature requested there is unfeasible. We cannot
> > disable optimistic locking with a switch. We would need an alternate
> > JAR
> > with no version columns declared in the mapping files.
> >
> > -Alejandro
> >
> > El mar, 03-02-2009 a las 16:30 +0000, trevor kirby escribió:
> >> Thanks for the branch and the work so far guys. The two issues
> >> mentioned below are now a blocker for the SOA CP because of NTT, so I
> >> would appreciate it if you could get the fixes into the 3.2.5 sp
> >> branch
> >> as quickly as you can. I've included Julian Coleman on the CC, can
> >> you
> >> make sure you include him on any emails about these JIRA's or NTT.
> >>
> >> Trev
> >>
> >>
> >>> Hi Alejandro,
> >>>
> >>> thanks for all you effort to keep the jbpm3 code base in prime shape
> >>>
> >>> http://jbpm.dyndns.org:8280/hudson/job/jBPM3-Container-Matrix/
> >>>
> >>> After a conversation with Trevor, I created a jbpm-3.2.5.SP1 branch
> >>> that is going to take the place of 3.2.4.SP1 in the SOA platform.
> >>> The
> >>> benefit is that we don't have to merge stuff back to 3.2.4.SP that
> >>> was
> >>> already fixed in 3.2.5.GA.
> >>>
> >>> I already merged the changes from 3.2.4.SP to 3.2.5.SP there there
> >>> is
> >>> (hopefully) nothing more for you to do.
> >>>
> >>> Trevor also mentioned the importance of NTT issues again.
> >>>
> >>> https://jira.jboss.org/jira/browse/JBPM-1952
> >>> https://jira.jboss.org/jira/browse/JBPM-1953
> >>>
> >>> Perhaps you could have a look at those next and if possible also
> >>> apply
> >>> the necessary changes to 3.2.5.SP?
> >>>
> >>> The jbpm-3.2.5.SP QA is here
> >>>
> >>> http://jbpm.dyndns.org:8180/hudson
> >>>
> >>> cheers
> >>> -thomas
> >>
> >
>
> ---
> Mark Little
> mlittle(a)redhat.com
>
> JBoss, a Division of Red Hat
> Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod
> Street, Windsor, Berkshire, SI4 1TE, United Kingdom.
> Registered in UK and Wales under Company Registration No. 3798903
> Directors: Michael Cunningham (USA), Charlie Peters (USA), Matt
> Parsons (USA) and Brendan Lane (Ireland).
>
>
>
16 years, 8 months