The problem is that ICU 4.8.1.1 ships with version 2011k of time zone information. Current as of this writing is 2012j. The first version after 2011k that fixes the Europe/Kiev problem is 2011m.
Upstream responded to the bug report with instructions on how to update the timezone data in ICU. I was able to follow those instructions and generate a new libicudata that does not exhibit the problem described in this bug report, but after I do that, there are test failures that look real; i.e., they do not appear to be cases where the test data is incorrect and the tests fail because data was corrected in the zone files. What I did: svn co http://source.icu-project.org/repos/icu/data/trunk/tzdata/icunew/2012j/44 zoneinfo-2012j Copy icudt48l.dat from source/data/in in the ICU source distribution to zoneinfo-2012j/le In that directory: for i in *.res; do icupkg -a $i icudt48l.dat; done The new icudt48l.dat file has the new time zone data in it. This is per directions here: http://userguide.icu-project.org/datetime/timezone#TOC-Updating-the-Time-Zone-Data# I could then replace icudt48l.dat in the source distribution with the new one by putting it in the debian directory, registering it in debian/source/include-binaries, and adding a post-patches target that overwrites the one in the sources with the new one. However, in the resulting build, there are automated test failures that do not appear in the original build. If I try updating the zone information to the closest version that contains the fix for Europe/Kiev, which is 2011m, I still get test failures after replacing the data file. In addition to this problem, I have to include a binary file in the debian directory that I generate by hand. A simple patch to zoneinfo64.txt won't do it because that data is not regenerated by default. If I delete the data file and try to make the build regenerate it, the build fails. This means that I can't generate a clean patch that only addresses the problem in the bug report, and even if I could, the change causes other test failures. The bottom line is that I don't believe I can update this at this point in the release process with confidence that I am not going to introduce other problems, and I don't think I can, in good conscience, ask the release team to make a freeze exception. I will leave this bug open in case anyone else is able to take it further. I know this is a significant issue and that it's taken a long time to get to this point without a satisfactory resolution, and I am sorry for that, but I don't think there's much else I can do about it right now. After wheezy is released, I will upload the latest ICU, and once it transitions to testing (which will be a while, I imagine), it can be backported to wheezy. ICU itself has few dependencies, so it should be pretty easy to backport. -- Jay Berkenbilt <q...@debian.org> -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
