[gnome-love] Re: GNOME Lovers Needed: l10n work for locations database

[Roozbeh Pournader <roozbeh farsiweb info>]

On Mon, 2004-11-29 at 22:40, Davyd Madeley wrote:
> gweather ships a very comprehensive locations database. GNOME-Applets
> 2.9.2 has some 3000+ more locations to this database[1] from syncing
> with the latest METAR list, but now we need your help.

As I previously mentioned, GNU miscfiles is also planning to do i18n
work on a certain list of locations, specially airports with 3-letter
IATA codes (gweather uses ICAO codes). We are far from 3000 locations,
of course, and the i18n work has not yet started either. But I believe
we've been good in spelling European names correctly and having the
proper and full names of the airports recorded in many cases. It still
needs to be proofread, but to get some idea of the current data, you can
check it at
<http://savannah.gnu.org/cgi-bin/viewcvs/*checkout*/miscfiles/miscfiles/airport?rev=HEAD&content-type=text/plain>.

I've been personally involved with growing the list, which includes
using free sources, so I have some idea on how this works. You should
even follow the news, sometimes...

BTW, may I ask where can one fetch the METAR list you mentioned?

> The database now has support for <city> tags (as well as <region>,
> <country> and <state>). While region, country, and state are all easy to
> work out, city tags are hard, it's easy to figure out what locations are
> in a city (sometimes) but not what their local names are. As a result,
> many names are things like "Location 1" or "My Location (alternate)".
>
> This is where you come in, spread around the world, and having a good
> knowledge of local, state and maybe even national geography, you will be
> able to clean up the locations file for your area.
> [...]

Such large lists of non-code need good maintenance, which may be hard or
impossible to do using the current bugzilla+cvs methodology. The only
project of the kind I know which has been working good enough, is the
English Wikipedia, which makes me believe the best option for doing such
a thing is using some wiki-like technology.

> Remember that all strings in the Locations.xml.in file should be in the
> C locale (US english). This means that locations should use their
> American names or anglicized names. Otherwise use the local name. If you
> use an anglicized name, add a comment to the XML so that translators
> know about the local name.

Anglicized names are not necessarily obvious. When you mention that, you
should mention your source. I have been using the preferences of the
English Wikipedia and the recommendations of The Oxford Style Guide
(R M Ritter, Oxford University Press, 2002), but I really believe
different US style guides will recommend different things. It would be
good if you could find one which has some good recommendations on city
naming and use (and recommend) that.

> Also, maintaining the locations database has been proving to be an
> increasingly time consuming task for me. If you feel you might be up to
> scratch at maintaining this database, drop me a line, I could do with
> some help.

You're taking an enormous project. If you do not wish to think about
changing the maintenance technology, I would recommend breaking it into
multiple files and delegating different parts to different people, based
on the area. Fortunately, ICAO codes are already sorted according to the
geographical area (codes starting with "O" are for Middle East and Asia,
those with "OI" are in Iran, etc.), so that can be used as a good idea
how to draw the lines between the files.

Also, I would like to ask if anyone is able to contribute a certain
amount of money to make it possible to buy ICAO and IATA publications on
these codes and how they are defined or updated?

roozbeh