[Infowarrior] - Time-zone database down

[Richard Forno]

(c/o CG)

Time-zone database down

http://blog.joda.org/2011/10/today-time-zone-database-was-closed.html?m=1

Today, the time-zone database was closed down.

It is perhaps easy to read that line, think it doesn't affect you, and then move on. But thats just not the case.

The time-zone database (sometimes referred to as the Olson database) is the computing world's principle source of time-zone data. It is embedded in every Unix and Java for starters, and will be used by many websites and probably by your iPhone. You may know it via the IDs, such as "Europe/London" or "America/New_York".

But, perhaps you're thinking that time-zones don't change? Well that may be true for America and the EU right now, but certainly isn't for the rest of the world. Governments change their time-zones all the time, and the decisions are frequently very political. I'd estimate there are between 20 and 100 separate changes made around the globe each year. And these can be at very short notice, triggered by earthquakes for example.

The time-zone database tracks all this information and creates a standard format file that describes it. I would show you an example of the file, but then perhaps I'd be sued....

The database itself was run as an open source project, led by Arthur David Olson, supported by many others. The data was published as a set of files about 15 times a year, and then picked up by users everywhere.

The complaint itself comes from Astrolabe, Inc, whose website looks like a company I would avoid doing business with.

The complaint is that Astrolabe produce a work, the "ACS Atlas", which is referenced by the time-zone database (some sources suggest that Astrolabe may have recently purchased the work). Astrolabe claim copyright over their work and thus believe that the time-zone databse should not have released their information to the public domain. The case is targetted at two private individuals - Arthur David Olson and Paul Eggert, who have hosted the website for many years.

The key passage in the time-zone databse files is this:

# From Paul Eggert (2006-03-22):
# A good source for time zone historical data in the US is
# Thomas G. Shanks, The American Atlas (5th edition),
# San Diego: ACS Publications, Inc. (1991).
# Make sure you have the errata sheet; the book is somewhat useless without it.
# It is the source for most of the pre-1991 US entries below.

For obvious reasons, I'll refrain on commenting on the rights and wrongs of the case, although I will note that facts like the phonebook cannot be copyrighted. A detailed response from one site taken down is now available. Instead I'll focus on the impact.

The impact of this is severe for anyone that uses it - whether via Java, Unix or some other means. This really is the key tool used by everyone to tell the right time globally. We all owe a debt of gratitude to the database maintainers who have worked on this for many, many years at zero cost to the industry and for zero financial gain.

So, right now the global situation is that there is no longer a single central location for time-zone information for computing. I'm sure that each major user project (like the Unix distros) will patch their own versions as best they can, but the stricter ones might argue that the current data is tainted and want to remove even that. This could get very messy very quickly.

Both Joda-Time and ThreeTen/JSR-310 use the data to build timezone information. ThreeTen/JSR-310 in particular provides this information in huge detail to applications. The worst case scenario is that multiple groups start up to provide this data in the future, and applications are then responsible for handling multiple competing data sources.

This data is so key to the world at this point that it needs to be formalised and run by a group with more legal and financial backing. Efforts had been ongoing to achieve this, but they may now be in jeopardy - who would want to take on a project being legally attacked?.

I hereby call on the industry leaders to help sort this out - IBM, Oracle, Apple, Google, RedHat I'm looking at you.

Update: I didn't include Microsoft here because Windows has its own time-zone data files.

In the meantime, could I please ask that anyone thinking of patching the data on a temporary basis, or trying to recreate it from scratch, re-uses the existing file format. There is no reason to believe that the C code or file format is tainted by the lawsuit, just the data. So, lets all please try to minimise the mess that could happen if everyone starts to go their own way.