How to figure out which, if any, update changed /etc/localtime from a link to a hard file?

[David M. Karr]

NoOp wrote:
> On 04/16/2009 10:29 PM, David M. Karr wrote:
>> Brian McKee wrote:
>>> On Thu, Apr 16, 2009 at 5:24 PM, David M. Karr
>>> <davidmichaelkarr at gmail.com> wrote:
>>>> So, I had recently discovered a bug with Java not using the correct date
>>>> ranges for DST.  What I've determined is that Java requires
>>>> /etc/localtime to be a symlink, not a hard file.  When I replaced the
>>>> hard file with a link to the correct file, it fixed my problem.
>>>>
>>>> Now, what I'd like to find out is how and when /etc/localtime became a
>>>> hard file.
>>>>
>>>> I have no idea whether the original Ubuntu 8.10 installation was in this
>>>> state or not.  It could have happened when I installed one of the many
>>>> updates that come through.
>>>>
>>>> is there any way to search through the list of updates that I've
>>>> installed to see whether any of them would have done this?
>>> Huh.  I thought java had it's own tzdata stuff and didn't use the
>>> system.  Perhaps that's changed.
>>>
>>> I'm guessing the packages you are referring to are tzdata and
>>> tzdata-java,  but I would have thought breaking java's timezone would
>>> be a big bug lots of people would have noticed by now.  Launchpad show
>>> anything?
>>>
>>> Brian
>>>
>> Now that I know what to search for, I found the following, which has a 
>> lot of history on the problem: 
>> <https://bugs.launchpad.net/ubuntu/+source/sun-java5/+bug/49068>.
>>
> 
> Which version of tzdata-java do you have installed?
> 
> http://packages.ubuntu.com/search?searchon=names&keywords=tzdata
> http://packages.ubuntu.com/intrepid-updates/tzdata-java
> https://launchpad.net/ubuntu/+source/tzdata

I have version 2009a.

> The most recent updata for intrepid was:
> 
> 2009f-0ubuntu0.8.10
> Published in intrepid-proposed on 2009-04-16
> tzdata (2009f-0ubuntu0.8.10) intrepid-proposed; urgency=low
>   * New timezone data 2009f: (LP: #358232):
>     - Fix DST rules for Jordan, Palestine, and Pakistan.
>  -- Martin Pitt <email address hidden>   Thu, 16 Apr 2009 12:41:46 +0200

That was, like, yesterday. :)

I don't see that it matters, however.  It seems pretty clear the tzdata 
for my timezone is perfectly fine, but the algorithm used to find the 
data was broken by the fact that /etc/localtime wasn't a symlink.  I 
verified that fixing this problem fixed the problem on multiple JVMs, 
even ones that I hadn't even run the 2009a tzdata on.