Build failed in Jenkins: ManifoldCF » ManifoldCF-mvn #11

[Apache Jenkins Server]

See 


Changes:

[Karl Wright] Prepare trunk for 2.19 development

[Karl Wright] Tie off release 2.18

[Karl Wright] Get JDK 11 building this time for real

[Karl Wright] Refactor tabs in connector-build

[Karl Wright] Include jaxb so that cmis test works on jdk 11

[Karl Wright] Get trunk to build on jdk11 with cxf connector


--
[...truncated 1.94 MB...]
[ERROR]   protected static void writeField(ModifiableSolrParams out, String 
fieldName, List fieldValues)
[ERROR] ^
[ERROR] 
:899:
 warning: no @param for out
[ERROR]   protected static void writeField(ModifiableSolrParams out, String 
fieldName, String fieldValue)
[ERROR] ^
[ERROR] 
:899:
 warning: no @param for fieldName
[ERROR]   protected static void writeField(ModifiableSolrParams out, String 
fieldName, String fieldValue)
[ERROR] ^
[ERROR] 
:899:
 warning: no @param for fieldValue
[ERROR]   protected static void writeField(ModifiableSolrParams out, String 
fieldName, String fieldValue)
[ERROR] ^
[ERROR] 
:905:
 warning: no @param for out
[ERROR]   protected void writeACLs(ModifiableSolrParams out, String aclType, 
String[] acl, String[] denyAcl)
[ERROR]  ^
[ERROR] 
:905:
 warning: no @param for aclType
[ERROR]   protected void writeACLs(ModifiableSolrParams out, String aclType, 
String[] acl, String[] denyAcl)
[ERROR]  ^
[ERROR] 
:905:
 warning: no @param for acl
[ERROR]   protected void writeACLs(ModifiableSolrParams out, String aclType, 
String[] acl, String[] denyAcl)
[ERROR]  ^
[ERROR] 
:905:
 warning: no @param for denyAcl
[ERROR]   protected void writeACLs(ModifiableSolrParams out, String aclType, 
String[] acl, String[] denyAcl)
[ERROR]  ^
[ERROR] 
:922:
 warning: no @param for inputDoc
[ERROR]   protected void writeACLsInSolrDoc( SolrInputDocument inputDoc, String 
aclType, String[] acl, String[] denyAcl )
[ERROR]  ^
[ERROR] 
:922:
 warning: no @param for aclType
[ERROR]   protected void writeACLsInSolrDoc( SolrInputDocument inputDoc, String 
aclType, String[] acl, String[] denyAcl )
[ERROR]  ^
[ERROR] 
:922:
 warning: no @param for acl
[ERROR]   protected void writeACLsInSolrDoc( SolrInputDocument inputDoc, String 
aclType, String[] acl, String[] denyAcl )
[ERROR]  ^
[ERROR] 
:922:
 warning: no @param for denyAcl
[ERROR]   protected void writeACLsInSolrDoc( SolrInputDocument inputDoc, String 
aclType, String[] acl, String[] denyAcl )
[ERROR]  ^
[ERROR] 
:1751:
 warning: no @param for inputField
[ERROR]   protected static String makeSafeLuceneField(String inputField)
[ERROR]   ^
[ERROR] 

Jenkins build is back to normal : ManifoldCF » ManifoldCF-Artifacts-Ant-JDK11 #12

[Apache Jenkins Server]

See 

Re: JDK 11

[Karl Wright]

Thanks for the comment.

OpenJDK, though, has also changed.  It always has been released from a fork
of the same java that goes into the Oracle releases.  While it might be
maintained by a different bunch of people, strategic decisions made by
Oracle also impact OpenJDK and, under the license terms, I cannot imagine
the situation changing.

Karl


On Sat, Dec 12, 2020 at 7:20 AM Piergiorgio Lucidi 
wrote:

> Hi Karl,
>
> I think that we should consider only OpenJDK edition of JDK 11 without
> considering the Oracle version.
>
> Probably we could be lucky with OpenJDK, anyway we have to go in deep also
> with OpenJDK investing more time on this.
>
> Hope to find some time soon to understand more details for this porting
> task to do :-P
>
> PJ
>
> Il Sab 12 Dic 2020, 13:01 Karl Wright  ha scritto:
>
> > Hi,
> >
> > Whereas I was under the impression that one of our committers had
> addressed
> > the JDK 11 problem in ManifoldCF, upon detailed inspection and trial I
> have
> > determined that this is in fact quite false.  The reason is that the
> > relatively new csws connector used a ton of J2EE classes that are no
> longer
> > bundled with JDK 11.
> >
> > I've managed to get this to build now, while adding a number of these
> > classes to the appropriate classpaths in connector-build.xml, but here's
> > the problem: unless somebody actually tries this connector, and uses it
> > under JDK 11 in a Livelink environment, I have no idea what classes may
> be
> > missing at runtime.  So I am hoping we have a committer somewhere who
> might
> > be able to work with me on this experiment.
> >
> > There's more news, and it isn't good: Oracle has deprecated some
> ubiquitous
> > methods and some techniques that many open-source projects we depend on
> > use.  For example, new Long(long) is going away.  I have no idea why they
> > did this but they must know it will effectively deprecate 100% of the
> Java
> > codebase that is lightly maintained, and force massive rework of pretty
> > near all opensource code.  That MUST be intentional on Oracle's part, and
> > to me it represents the writing on the wall concerning the lifetime of
> > legacy ManifoldCF connectors.  It will likely not be possible to support
> > our existing connector family when this happens.
> >
> > Meanwhile, we still need to go through the JDK 11 deprecations in our own
> > codebase and fix those, as well as verify proper function of all
> connectors
> > on JDK 11.  I'll be tackling that project myself in my voluminous spare
> > time.
> >
> > Karl
> >
>

Re: JDK 11

[Piergiorgio Lucidi]

Hi Karl,

I think that we should consider only OpenJDK edition of JDK 11 without
considering the Oracle version.

Probably we could be lucky with OpenJDK, anyway we have to go in deep also
with OpenJDK investing more time on this.

Hope to find some time soon to understand more details for this porting
task to do :-P

PJ

Il Sab 12 Dic 2020, 13:01 Karl Wright  ha scritto:

> Hi,
>
> Whereas I was under the impression that one of our committers had addressed
> the JDK 11 problem in ManifoldCF, upon detailed inspection and trial I have
> determined that this is in fact quite false.  The reason is that the
> relatively new csws connector used a ton of J2EE classes that are no longer
> bundled with JDK 11.
>
> I've managed to get this to build now, while adding a number of these
> classes to the appropriate classpaths in connector-build.xml, but here's
> the problem: unless somebody actually tries this connector, and uses it
> under JDK 11 in a Livelink environment, I have no idea what classes may be
> missing at runtime.  So I am hoping we have a committer somewhere who might
> be able to work with me on this experiment.
>
> There's more news, and it isn't good: Oracle has deprecated some ubiquitous
> methods and some techniques that many open-source projects we depend on
> use.  For example, new Long(long) is going away.  I have no idea why they
> did this but they must know it will effectively deprecate 100% of the Java
> codebase that is lightly maintained, and force massive rework of pretty
> near all opensource code.  That MUST be intentional on Oracle's part, and
> to me it represents the writing on the wall concerning the lifetime of
> legacy ManifoldCF connectors.  It will likely not be possible to support
> our existing connector family when this happens.
>
> Meanwhile, we still need to go through the JDK 11 deprecations in our own
> codebase and fix those, as well as verify proper function of all connectors
> on JDK 11.  I'll be tackling that project myself in my voluminous spare
> time.
>
> Karl
>

JDK 11

[Karl Wright]

Hi,

Whereas I was under the impression that one of our committers had addressed
the JDK 11 problem in ManifoldCF, upon detailed inspection and trial I have
determined that this is in fact quite false.  The reason is that the
relatively new csws connector used a ton of J2EE classes that are no longer
bundled with JDK 11.

I've managed to get this to build now, while adding a number of these
classes to the appropriate classpaths in connector-build.xml, but here's
the problem: unless somebody actually tries this connector, and uses it
under JDK 11 in a Livelink environment, I have no idea what classes may be
missing at runtime.  So I am hoping we have a committer somewhere who might
be able to work with me on this experiment.

There's more news, and it isn't good: Oracle has deprecated some ubiquitous
methods and some techniques that many open-source projects we depend on
use.  For example, new Long(long) is going away.  I have no idea why they
did this but they must know it will effectively deprecate 100% of the Java
codebase that is lightly maintained, and force massive rework of pretty
near all opensource code.  That MUST be intentional on Oracle's part, and
to me it represents the writing on the wall concerning the lifetime of
legacy ManifoldCF connectors.  It will likely not be possible to support
our existing connector family when this happens.

Meanwhile, we still need to go through the JDK 11 deprecations in our own
codebase and fix those, as well as verify proper function of all connectors
on JDK 11.  I'll be tackling that project myself in my voluminous spare
time.

Karl