Community
Participate
Working Groups
Just now thought to look at "comparator output" and saw there wasn't any. In one of the obscure error logs is this message: BUILD FAILED /shared/eclipse/builds/4I/siteDir/eclipse/downloads/drops4/I20131126-0800/eclipse.platform.releng.aggregator/eclipse.platform.releng.tychoeclipsebuilder/eclipse/buildScripts/eclipse_compare.xml:14: Property orbit-repo.url was circularly defined. My guess is we removed the main "orbit-repo.url" variable from parent pom ... but perhaps something else depends on it?
The source of the problem is technically unrelated to "comparator" ... just so happens to occur while we try to do that task. The "hint" is looking at line 14, of eclipse_compare.xml <loadproperties srcfile="${buildDirectory}/${buildId}/mavenproperties.properties" /> The "mavenproperties.properties" file is one we deliberately "write out" in the parent pom, to provide some persistence of the value of variables used in the build ... and in some cases, we do use those values else where ... in this case, we actually want the "comparator.repo" value ... just to print out in the comparator reports. But, looking at that mavenproperties.properties file from that I-build I see (in part) 27 comparator.repo=http://download.eclipse.org/eclipse/updates/4.4-I-builds 28 29 orbit-repo.url=${orbit-repo.url} 30 aspectj-repo.url=http://download.eclipse.org/tools/ajdt/aspectj/update/e42 Hence (line 29) the "circular reference" is a side effect of us removing orbit-repo.url property (and value) from the parent pom.
Fixed problem in master and R4_3_maintenance. http://git.eclipse.org/c/platform/eclipse.platform.releng.aggregator.git/commit/?h=R4_3_maintenance&id=8b9fd061d6fec3886b28825b9d1b3f4de541b432 I'll see if I can generate "by hand" and republish.
I was able to "generate by hand" and rsync up to downloads ... http://download.eclipse.org/eclipse/downloads/drops4/I20131126-0800/buildlogs/comparatorlogs/buildtimeComparatorUnanticipated.log.txt and seems some "feature.xml" files are listed there. I'll need to look closer to see if anything to be worry about or not. (I've seen a case or two where the "download size" changed by 1 byte, for some reason). Could be something serious, but I'd think if "way off" the junit tests wouldn't run ... so ... we'll see .... tomorrow.
(In reply to David Williams from comment #3) > ... so ... we'll see .... tomorrow. Well, I couldn't rest until I took a quick peek ... it is one of those "size" related things ... not sure what causes that .. and didn't even get enough context to know which plugins are different sizes ... but doesn't seem anything to worry about ... maybe some difference in "packing"? $ diff /home/data/users/david_williams/temp/p2.sdk/feature.xml feature.xml 40,41c40,41 < download-size="66" < install-size="138" --- > download-size="72" > install-size="149" $ diff feature.xml /home/data/users/david_williams/temp/p2.core/feature.xml 171,172c171,172 < download-size="72" < install-size="149" --- > download-size="66" > install-size="138"
