You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@uima.apache.org by Richard Eckart de Castilho <re...@apache.org> on 2020/04/19 15:56:19 UTC

[VOTE] Release UIMA uimaFIT 3.1.0 RC2

Hi,

I've posted the uimaFIT 3.1.0 release candidate 2.

uimaFIT 3.1.0 is a minor feature and bugfix release. On supported platforms, it serves mostly as 
a drop-in replacement for previous uimaFIT 3.x versions. However, due to a clash in method
signatures, you may notice incompatible changes in the methods provided ExternalResourceFactory.
For details, please refer to the migration section in the documentation.

Changes since RC1:

- UIMA-6222 - XML-related unit test fails on Java 9+
- UIMA-6223 - Groovy plugin causes warning during build
- UIMA-6224 - uimaFIT does not build on Java 14
- Without issue: commented out various print statements in unit tests to reduce log clutter during build

Changes:
https://issues.apache.org/jira/secure/ReleaseNote.jspa?projectId=12310570&version=12343882

The source and binary tar/zips are here:
https://dist.apache.org/repos/dist/dev/uima/uimafit/3.1.0-rc2/

The Maven artifacts are here:
https://repository.apache.org/content/repositories/orgapacheuima-1254

The GIT tag is here:
https://github.com/apache/uima-uimafit/tree/uimafit-3.1.0

Clone using: git clone -b uimafit-3.1.0 https://github.com/apache/uima-uimafit.git

See http://uima.apache.org/testing-builds.html for suggestions on how to test
release candidates.

Please vote on release:

[ ] +1 OK to release
[ ] 0   Don't care
[ ] -1 Not OK to release, because ...

Thanks.

-- Richard

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Marshall Schor <ms...@schor.com>.

perfect...

On 5/4/2020 1:07 PM, Richard Eckart de Castilho wrote:
> On 4. May 2020, at 19:00, Marshall Schor <ms...@schor.com> wrote:
>> It's marked unresolved / open.  Is the version number wrong for 3.1.0?
> I was considering to cancel the vote and add this fix to 3.1.0, but then
> decided against it and forgot to update the fix version. It is now set
> to 3.2.0.
>
> -- Richard

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Richard Eckart de Castilho <re...@apache.org>.

On 4. May 2020, at 19:00, Marshall Schor <ms...@schor.com> wrote:
> 
> It's marked unresolved / open.  Is the version number wrong for 3.1.0?

I was considering to cancel the vote and add this fix to 3.1.0, but then
decided against it and forgot to update the fix version. It is now set
to 3.2.0.

-- Richard

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Marshall Schor <ms...@schor.com>.

issue UIMA-6226 https://issues.apache.org/jira/browse/UIMA-6226
is marked as belonging to 2.6.0 uima fit and ** uima 3.1.0 **.

It's marked unresolved / open.  Is the version number wrong for 3.1.0?

-Marshall

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Marshall Schor <sc...@apache.org>.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512
 

signatures - OK
compare source-release with git tag: OK
build-from-source (using Java 14) OK with following noted:
   JCasUtilv3Test: some tests returning NaN or Infinity as speed-up-factor
   JCasUtilv3Test: some test report < 1 speedup factor

   Javadoc warnings : missing @params
   japicmp-maven-plugin:0.14.3:  illegal reflective access by groovy
   maven-plugin-plugin: @Component deprecated, replace with @Parameter:
o.a.u.fit.maven.EnhanceMojo, and GenerateDescriptorsMojo
   maven-assembly-plugin: warning, untriggered exclusion filter for
aopalliance:aopalliance

   some instances of "No NOTICE file was found - content not loaded into
property [postNoticeText]" e.g. uimafit-cpe
API Change Report: OK
    asked about semantic version
IssuesFixed OK
README OK
spot check licenses OK

[x] +1 OK to release

- -Marshall Schor
-----BEGIN PGP SIGNATURE-----
 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=V6s4
-----END PGP SIGNATURE-----

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Richard Eckart de Castilho <re...@apache.org>.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

* Build from sources zip (using Java 8): 			OK
* Build from sources zip (using Java 14): 			OK
* Checked README: 						OK
* Checked fixed issues: 					OK
* Checked API changes report: 					OK
* Checked documentation in relation to API changes:             OK
* Compared git tag vs sources ZIP: 				OK
* Build DKPro Core against RC:					OK
* Checked NOTICE and LICENSE files in the binary release: 	OK
 Note: there was an update of Spring which affected these

[X] +1 OK to release

Richard
-----BEGIN PGP SIGNATURE-----
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=xtUa
-----END PGP SIGNATURE-----

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Marshall Schor <ms...@schor.com>.

while testing (with Java14) noticed these in the build-from-source (non are
blockers):

   JCasUtilv3Test: some tests returning NaN or Infinity as speed-up-factor
   JCasUtilv3Test: some test report < 1 speedup factor

   groovy-maven-plugin: reports sometimes that
   Javadoc warnings : missing @params
    japicmp-maven-plugin:0.14.3:  illegal reflective access by groovy

    maven-plugin-plugin: @Component deprecated, replace with @Parameter:
o.a.u.fit.maven.EnhanceMojo, and GenerateDescriptorsMojo
    maven-assembly-plugin: warning, untriggered exclusion filter for
aopalliance:aopalliance

    some instances of "No NOTICE file was found - content not loaded into
property [postNoticeText]" e.g. uimafit-cpe

Testing continuing ... -Marshall

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Richard Eckart de Castilho <re...@apache.org>.

On 4. May 2020, at 18:48, Richard Eckart de Castilho <re...@apache.org> wrote:
> 
>> the uimafit-parent pom has no specs on backwards compatibility. Known and intended behaviour?

And yes: there is no forced check that the semantic versioning rules are abided by at the moment.

Cf. the mails with Marshall wrt. naming the new version.

-- Richard

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Richard Eckart de Castilho <re...@apache.org>.

Hi,

> On 4. May 2020, at 16:55, Viorel Morari <vi...@averbis.com> wrote:
> 
> the uimafit-parent pom has no specs on backwards compatibility. Known and intended behaviour?

The uimafit-parent POM defines this property

  <api_check_oldVersion>3.0.0</api_check_oldVersion>

That controls against which version the "japicmp-maven-plugin" inherited from the UIMA parent POM
runs the comparison. If you check the effective POM e.g. of the uimafit-core module, you'll see
this:

      <plugin>
        <groupId>com.github.siom79.japicmp</groupId>
        <artifactId>japicmp-maven-plugin</artifactId>
        <version>0.14.3</version>
        <executions>
          <execution>
            <phase>verify</phase>
            <goals>
              <goal>cmp</goal>
            </goals>
            <configuration>
              <parameter>
                <onlyModified>true</onlyModified>
                <postAnalysisScript>/Users/bluefire/git/uima-uimafit/uimafit-core/../uimafit-parent/src/main/groovy/api-report.groovy</postAnalysisScript>
              </parameter>
              <oldVersion>
                <dependency>
                  <groupId>org.apache.uima</groupId>
                  <artifactId>uimafit-core</artifactId>
                  <version>3.0.0</version>
                </dependency>
              </oldVersion>
            </configuration>
          </execution>
        </executions>
      </plugin>

So here you can see that the API of the uimafit-core 3.1.0 is compared against the 3.0.0 version.

Cheers,

-- Richard

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Viorel Morari <vi...@averbis.com>.

Hi,

the uimafit-parent pom has no specs on backwards compatibility. Known and intended behaviour?


-Viorel

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2 - semantic version?

Posted by Marshall Schor <ms...@schor.com>.

I agree, makes sense :-) -Marshall

On 5/4/2020 12:41 PM, Richard Eckart de Castilho wrote:
> Hi,
>
>> On 4. May 2020, at 18:30, Marshall Schor <ms...@schor.com> wrote:
>>
>> API Change Report shows some "bindResource" methods were removed.
>>
>> Sometimes, removing public methods users (may have) depended on implies incrementing
>> the major component of the version.
> Yes, that is correct. Normally, one would increment the major version.
>
>> I know there have been some iterations around fixing these methods, so it may be
>> fine to not increment the major number for this release.
>>
>> What do you think?
> I have tried to keep the breaking changes to the public API minimal, mostly deprecating
> stuff and only in a few cases where breaking changes could not be avoided due to "bugs"
> in the API design, make these breaking changes. 
>
> Anybody "depending" on the broken methods may well have done so accidentally and I'd
> rather they'd be force to review their code than continue being lucky or running 
> around with broken code they might not even know about.
>
> Therefore, although it is not following exactly the rules of the semantic versioning,
> I'd tend to increase only the middle version.
>
> -- Richard
>

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2 - semantic version?

Posted by Richard Eckart de Castilho <re...@apache.org>.

Hi,

> On 4. May 2020, at 18:30, Marshall Schor <ms...@schor.com> wrote:
> 
> API Change Report shows some "bindResource" methods were removed.
> 
> Sometimes, removing public methods users (may have) depended on implies incrementing
> the major component of the version.

Yes, that is correct. Normally, one would increment the major version.

> I know there have been some iterations around fixing these methods, so it may be
> fine to not increment the major number for this release.
> 
> What do you think?

I have tried to keep the breaking changes to the public API minimal, mostly deprecating
stuff and only in a few cases where breaking changes could not be avoided due to "bugs"
in the API design, make these breaking changes. 

Anybody "depending" on the broken methods may well have done so accidentally and I'd
rather they'd be force to review their code than continue being lucky or running 
around with broken code they might not even know about.

Therefore, although it is not following exactly the rules of the semantic versioning,
I'd tend to increase only the middle version.

-- Richard

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2 - semantic version?

Posted by Marshall Schor <ms...@schor.com>.

API Change Report shows some "bindResource" methods were removed.

Sometimes, removing public methods users (may have) depended on implies incrementing
the major component of the version.

I know there have been some iterations around fixing these methods, so it may be
fine to not increment the major number for this release.

What do you think?

-Marshall

[RESULT][VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Richard Eckart de Castilho <re...@apache.org>.

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Viorel Morari <vi...@averbis.com>.

compared git tag with source-release ............................. OK
deleted .m2 built files, built from source (with Java 8) ......... OK
spot check DEPENDENCIES .......................................... OK
spot check LICENSE/NOTICE/README ................................. OK
check parent-pom settings for API compatibility .................. OK
documentation .................................................... OK

[x] +1 OK to release

- Viorel

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Marshall Schor <ms...@schor.com>.

good to ping, as I've been swamped with work... I'll try to get to this soon :-)

On 4/29/2020 6:39 AM, Richard Eckart de Castilho wrote:
>> On 19. Apr 2020, at 17:56, Richard Eckart de Castilho <re...@apache.org> wrote:
>>
>> I've posted the uimaFIT 3.1.0 release candidate 2.
> Ping? ;)
>
> -- Richard

Re: [VOTE] Release UIMA uimaFIT 3.1.0 RC2

Posted by Richard Eckart de Castilho <re...@apache.org>.

> On 19. Apr 2020, at 17:56, Richard Eckart de Castilho <re...@apache.org> wrote:
> 
> I've posted the uimaFIT 3.1.0 release candidate 2.

Ping? ;)

-- Richard