Colophon

The tools behind the Journal

Penguin Hacker - courtesy of http://www.animationfactory.com Here are the tools we use to produce the journal, along with some other technical details that nobody really cares about.

French translation

Translating is easy. What is hard is keeping the translation of a Web site up-to-date with its original version. Indeed, contrary to a book, a Web site (or technical documentation, or pretty much anything that isn't printed on paper or carved on stone, for that matter) is a moving target. On our Web site for example, it is typical that Narelle will change the home page several times in a single week. The naive translator, who would simply copy the HTML file into a new file and translate that, would quickly lose track of changes, until their translation were so out-of-date that they may as well start again from scratch.

Translating the files

Luckily, the problem has been addressed by the Debian internationalisation team: They have created a tool called po4a, which I have slightly extended to support XHTML, to help with maintaining translations of documentation.

The translating process counts three phases, only one of which requires manual intervention:

Once the original PO file is written, we move into maintainance mode: if an original, English page changes, po4a can simply update the translated document using the same PO catalog. If actual content (as opposed to page layout) has changed, it'll mark those parts of the PO file as in need of a review. The translator then just updates what has changed, while the rest of the translation is kept intact. The translated pages can be generated from the catalog and the original pages.

Po4a is a tool written in Perl, based in GNU gettext. It should work pretty much out of the box on any Unix system. If you fetch po4a directly from their CVS, you'll also get the XHTML module I contributed.

Server setup

Now that we know how to produce translated documents, we need to set up the server so that translations are served as required. Normally, a user's browser is configured to request pages in a specific language (e.g. all Internet Explorers sold in Germany request pages in German). The language requested is specified at the same time as the browser requests the file.

We use Apache. With Apache, basically all you need to do is switch the MultiViews option on, and when asked for a file that doesn't exist, Apache will look up similar file names that would contain the language code specified in the request. For example, there is no 'index.shtml' file on our Photojournal. So when a browser requests that file, specifying French, Apache finds 'index.shtml.en' and 'index.shtml.fr' and chooses the latter.

Combined with the previous tool, that's all we need to keep our Web site up-to-date. We create PO files, we get po4a to generate files ending in '.fr' from files ending in '.en' and Apache picks the appropriate file.

Photography

We're currently taking most of our photos with a Fuji Finepix S5000 digital camera.

Most of the pictures taken after Christmas 2000 and up until May 2004 are taken with an old 35mm Chinon CM5. Older pictures are usually taken with random disposable cameras. We don't use any specific brand of film, usually the widely available Kodak Gold and Fujifilm Superia, in 200 or 400ASA.

After trying many labs around Edinburgh, we settled for Boots, which provides the best hardcopies. We also tried a couple of Kodak shops, but all the hardcopies were slightly blurred. Bonusprint provides an interesting, and very convenient, mail order developing service, however one of the first rolls we sent them came back with spots all over the copies, so we gave them up as well. So Boots it is, so far.

The oldest pictures are scanned from hard copies, using whatever scanner we could put our hands on. This being a long and cumbersome process, we then got many pictures developped straight to CD by Boots. Then Boots' services stopped being that interesting, so we bought a Microtek Filmscan 35, and did all the scanning at home directly from negatives. This proved to be a cheap and rather effective solution, however our scanning quality isn't as good as what Boots used to provide, and this still tends to be pretty slow. These days, we tend to still get Boot's CDs as it is much faster.

Computing

Post processing is mostly done on a Toshiba Satellite A30 laptop, running Microsft Windows XP. Picture processing is usually done with The Gimp.

HTML editing is done by hand using Vim. We are continually working towards W3C compliance. This goal is mostly reached since mid-2005, but for the occasionnal accident.

We get things working under Mozilla FireFox and final tests are done in the most up-to-date versions of Microsoft Internet Explorer 6, Opera and Lynx. If you have any gripe or serious problem with your favourite browser, please let us know and we'll see what we can do (if that browser is available for free, is W3C standard compliant and open source.)

Accessibility

Narelle specialises in making Web sites fully accessible regardless of user ability, browser type, Internet connection speed, screen resolution or physical disability.

This Web site is continually moving towards 100% accessibility.

Hosting

We are currently hosting the Web site ourselves. The server is a brand-less computer, with an Intel Celeron 2.50Ghz with 1GB of RAM. The Web server is Apache, running under Debian.

It is connected 24/7 to the Internet through our ADSL provider Free.fr.

We went through several hosts before:

Narelle and Yves' Photojournal