Bug 497030

Comment 1 David Williams 2016-06-29 13:30:03 EDT

I have realized the main reason for this is that we now use "job name" as part of what we look for in retrieving the data -- and those are partially hard code, such as in "publish.xml" we have statements such as 

    <property
      name="testsConfigExpected"
      value="ep46${buildType}-unit-lin64_linux.gtk.x86_64_8.0,ep46${buildType}-unit-mac64_macosx.cocoa.x86_64_8.0,ep46${buildType}-unit-win32_win32.win32.x86_8.0,ep46${buildType}-unit-cen64_linux.gtk.x86_64_8.0" />

We could probably make this work by changing 'ep46' to 'ep47' in master. 

BUT, a better solution would make use of some variables. 

We already have some "up stream" variables for ${eclipseStreamMajor} and ${eclipseStreamMinor} (and even ${eclipseStreamService}). For example, see 
genTestIndexes.xml. 

I think the 'publish.xml' will have access to those variables, by inheriting them from the calling scripts -- I think. (Sometimes hard to tell what paths these scripts take when called). 

But, I suggest we "just try it" and if does not work, we will find the places we need to make sure we "pass them" to the right places.

Comment 2 David Williams 2016-06-29 14:32:14 EDT

This is commit that fixes 4 files in master. 

http://git.eclipse.org/c/platform/eclipse.platform.releng.aggregator.git/commit/?id=37052431ce9229bb58777432f05b1b10e76fea7c

(Not all 4 are used, yet). 

I think "maintenance" might actually work, since the job still starts with "ep46" so not sure we need a fix on that branch. 

= = = = =

But besides *this* problem, a number of N-builds before we started making these changes were not being published (or, perhaps not running?) and I have not looked at what their problem was. But, this should help future N and I builds. 

= = = = = =

BTW, while it is possible to "republish" using master of these type of utilities, it is not especially easy, and I'd prefer not to take the time. So if anyone though important to get test results "this week" then probably easiest to "re-run" a build and its subsequent tests.

Comment 3 David Williams 2016-07-04 10:21:37 EDT

After looking at the reasons for bug 497207, I see the following error in the log, which is a result of the fix I choose for this *this* bug. 

Caused by: /shared/eclipse/builds/4N/siteDir/eclipse/downloads/drops4/N20160702-1500/eclipse.platform.releng.aggregator/eclipse.platform.releng.tychoeclipsebuilder/eclipse/helper.xml:37: The following error occurred while executing this line:
/shared/eclipse/builds/4N/siteDir/eclipse/downloads/drops4/N20160702-1500/eclipse.platform.releng.aggregator/eclipse.platform.releng.tychoeclipsebuilder/eclipse/helper.xml:99: The following error occurred while executing this line:
/shared/eclipse/builds/4N/siteDir/eclipse/downloads/drops4/N20160702-1500/eclipse.platform.releng.aggregator/eclipse.platform.releng.tychoeclipsebuilder/eclipse/buildScripts/publish.xml:229: This publish.xml script needs eclipseStreamMajor and eclipseStreamMinor
        at org.apache.tools.ant.ProjectHelper.addLocationToBuildException(ProjectHelper.java:568)

Comment 4 David Williams 2016-07-04 10:23:28 EDT

I will look for an easy way to "pass in" stream information, and if it is not easy, will resort back to hard coding "47" and leave the variable solution for future enhancement.

Comment 5 David Williams 2016-07-04 11:01:59 EDT

http://git.eclipse.org/c/platform/eclipse.platform.releng.aggregator.git/commit/?id=ad3245c3acc0e7e5d95efec76f57d1ab20943aee


I think fix was easy, if you accept "copy/paste" identical code as easy. 

We'll try that "quick and easy" fix to make sure that is the (only) problem, and in future someone can improve the code.