Thread: [Eclemma-jacoco-dev] Final preparations for release 0.5.3
Brought to you by:
mtnminds
|
From: Evgeny M. <man...@gm...> - 2011-07-02 11:39:20
|
Hi guys, I'd like to start release of JaCoCo 0.5.3 with publishing to Maven repository. But before doing so documentation must be updated and this is exactly what I'm doing now, but there is two issues and I need an advice : 1. Current Maven build behaves like Ant build - e.g. documentation contains qualified versions of libraries, but this is not always correct in case of jacoco-maven-plugin. E.g. current deployed documentation ( http://www.eclemma.org/jacoco/trunk/doc/maven.html ) mentions jacoco-maven-plugin version "0.5.3.20110621172127" , but it should mention "0.5.3-SNAPSHOT" in case of unreleased version and qualified version in case of released. Well, this can be simply solved by replacing "@qualified.bundle.version@" on "@project.version@" in documentation for jacoco-maven-plugin. But in this case Ant build wouldn't behave as Maven build anymore. And in this case our documentation will always contain reference on "SNAPSHOT" version, because we deploy documentation from trunk. So my suggestion is to remove substitution on jacoco-maven-plugin page and to have several sections - how to use snapshot version of plugin and how to use released version of plugin. 2. Maven build does not contain results of performance tests in documentation ( http://www.eclemma.org/jacoco/trunk/test/performance.txt ). And here's why : I've played with this tests. And from my point of view they completely useless, at least in documentation. Several executions of such tests in same environment will give us different results, so they are not stable. Moreover - results completely differ if I will make several executions using one JVM, or using different. So from developer point of view they useless, because they are not stable and can't provide comparison with previous release (to find degradations). And from user point of view they useless, because as a user I would expect not a generic numbers, but a real comparison with other coverage engines ( e.g. something like http://www.sonarsource.org/pick-your-code-coverage-tool-in-sonar-2-2/ ). So my suggestion Is to remove results of performance tests from documentation, at least for the moment. So WDYT? -- Best regards, Evgeny Mandrikov aka Godin <http://godin.net.ru> http://twitter.com/_godin_ |
|
From: Marc R. H. <hof...@mo...> - 2011-07-03 12:33:03
|
Hi Evgeny, I just run the Maven build locally and have to say that I'm absolutely impressed! From my point of view the Maven build should become the reference from now on. Therefore concerning 1) we should fix it to produce a proper result for the Maven build. Concerning 2) Since my machine was the only one to to create build I made the performance figures part of the build to see the figures evolve over time. I understand that this doesn't make sense any more when we run the build on different machines. So we can remove the link from documentation (and also the execution during build). I will find some solution for me to run it locally and keep the figures over time. Two remarks related to the release: Can you try to publish a snapshot build to maven and also to SF (download and website)? As soon as we switch to Maven this stream should also be established. When you do the release: Please add a new section in the changes.html first. Many thanks for this great work! Cheers, -marc On 02.07.11 13:38, Evgeny Mandrikov wrote: > Hi guys, > > I'd like to start release of JaCoCo 0.5.3 with publishing to Maven > repository. But before doing so documentation must be updated and this > is exactly what I'm doing now, but there is two issues and I need an > advice : > > 1. Current Maven build behaves like Ant build - e.g. documentation > contains qualified versions of libraries, but this is not always > correct in case of jacoco-maven-plugin. E.g. current deployed > documentation ( > http://www.eclemma.org/jacoco/trunk/doc/maven.html ) mentions > jacoco-maven-plugin version "0.5.3.20110621172127" , but it should > mention "0.5.3-SNAPSHOT" in case of unreleased version and > qualified version in case of released. Well, this can be simply > solved by replacing "@qualified.bundle.version@" on > "@project.version@" in documentation for jacoco-maven-plugin. > But in this case Ant build wouldn't behave as Maven build > anymore. And in this case our documentation will always contain > reference on "SNAPSHOT" version, because we deploy documentation > from trunk. So my suggestion is to remove substitution on > jacoco-maven-plugin page and to have several sections - how to use > snapshot version of plugin and how to use released version of plugin. > 2. Maven build does not contain results of performance tests in > documentation ( > http://www.eclemma.org/jacoco/trunk/test/performance.txt ). And > here's why : I've played with this tests. And from my point of > view they completely useless, at least in documentation. Several > executions of such tests in same environment will give us > different results, so they are not stable. Moreover - results > completely differ if I will make several executions using one JVM, > or using different. So from developer point of view they useless, > because they are not stable and can't provide comparison with > previous release (to find degradations). And from user point of > view they useless, because as a user I would expect not a generic > numbers, but a real comparison with other coverage engines ( e.g. > something like > http://www.sonarsource.org/pick-your-code-coverage-tool-in-sonar-2-2/ ). > So my suggestion Is to remove results of performance tests from > documentation, at least for the moment. > > So WDYT? > > -- > Best regards, > Evgeny Mandrikov aka Godin <http://godin.net.ru> > http://twitter.com/_godin_ > |
|
From: Evgeny M. <man...@gm...> - 2011-07-03 12:52:56
|
Hi Marc, See my comments below : On Sun, Jul 3, 2011 at 16:32, Marc R. Hoffmann <hof...@mo...>wrote: > Hi Evgeny, > > I just run the Maven build locally and have to say that I'm absolutely > impressed! From my point of view the Maven build should become the > reference from now on. > > Therefore concerning 1) we should fix it to produce a proper result for > the Maven build. > Ok - I'll do it. > Concerning 2) Since my machine was the only one to to create build I > made the performance figures part of the build to see the figures evolve > over time. I understand that this doesn't make sense any more when we > run the build on different machines. So we can remove the link from > documentation (and also the execution during build). I will find some > solution for me to run it locally and keep the figures over time. > Ok - I'll remove it. But in fact I would prefer if we will find a way to stabilize performance testing and a way to keep figures over time, so to make it usable by everybody. But I agree that this is not blocker for now and can wait. > Two remarks related to the release: > > Can you try to publish a snapshot build to maven and also to SF > (download and website)? As soon as we switch to Maven this stream should > also be established. > > When you do the release: Please add a new section in the changes.html > first. > Sure. My steps will be : 1. Finish documentation 2. Publish latest snapshot version into both locations - SF and Maven repository 3. Stage a Release 4. Verify staged version 5. Release into both locations - SF and Maven repository I will notify you, guys, on phases 2 and 4, so you can also participate in verification process ;) Many thanks for this great work! > > Cheers, > -marc > > > On 02.07.11 13:38, Evgeny Mandrikov wrote: > > Hi guys, > > > > I'd like to start release of JaCoCo 0.5.3 with publishing to Maven > > repository. But before doing so documentation must be updated and this > > is exactly what I'm doing now, but there is two issues and I need an > > advice : > > > > 1. Current Maven build behaves like Ant build - e.g. documentation > > contains qualified versions of libraries, but this is not always > > correct in case of jacoco-maven-plugin. E.g. current deployed > > documentation ( > > http://www.eclemma.org/jacoco/trunk/doc/maven.html ) mentions > > jacoco-maven-plugin version "0.5.3.20110621172127" , but it should > > mention "0.5.3-SNAPSHOT" in case of unreleased version and > > qualified version in case of released. Well, this can be simply > > solved by replacing "@qualified.bundle.version@" on > > "@project.version@" in documentation for jacoco-maven-plugin. > > But in this case Ant build wouldn't behave as Maven build > > anymore. And in this case our documentation will always contain > > reference on "SNAPSHOT" version, because we deploy documentation > > from trunk. So my suggestion is to remove substitution on > > jacoco-maven-plugin page and to have several sections - how to use > > snapshot version of plugin and how to use released version of > plugin. > > 2. Maven build does not contain results of performance tests in > > documentation ( > > http://www.eclemma.org/jacoco/trunk/test/performance.txt ). And > > here's why : I've played with this tests. And from my point of > > view they completely useless, at least in documentation. Several > > executions of such tests in same environment will give us > > different results, so they are not stable. Moreover - results > > completely differ if I will make several executions using one JVM, > > or using different. So from developer point of view they useless, > > because they are not stable and can't provide comparison with > > previous release (to find degradations). And from user point of > > view they useless, because as a user I would expect not a generic > > numbers, but a real comparison with other coverage engines ( e.g. > > something like > > > http://www.sonarsource.org/pick-your-code-coverage-tool-in-sonar-2-2/ ). > > So my suggestion Is to remove results of performance tests from > > documentation, at least for the moment. > > > > So WDYT? > > > > -- > > Best regards, > > Evgeny Mandrikov aka Godin <http://godin.net.ru> > > http://twitter.com/_godin_ > > > > > ------------------------------------------------------------------------------ > All of the data generated in your IT infrastructure is seriously valuable. > Why? It contains a definitive record of application performance, security > threats, fraudulent activity, and more. Splunk takes this data and makes > sense of it. IT sense. And common sense. > http://p.sf.net/sfu/splunk-d2d-c2 > _______________________________________________ > eclemma-jacoco-dev mailing list > ecl...@li... > https://lists.sourceforge.net/lists/listinfo/eclemma-jacoco-dev > -- Best regards, Evgeny Mandrikov aka Godin <http://godin.net.ru> http://twitter.com/_godin_ |