BarCampBlock : BarCampBlockSessionNotes

August 18, 2007

Convened by Peter Robinett

Talk based on one of Peter's pet peeves: Too many developers, particularly American ones, put off supporting other languages and display formats until they are firmly established in English. Adding support then can be hard. Just little differences can put people off: when living in London Facebook kept asking me to chose a state when entering an address. That is why copycat sites exist: they have the very importance difference for users of being in their local language and formats (regardless of what developers say about the functionality being the same).

It's not that hard if you consider other countries from the beginning. UTF-8 is present and well-supported in just about every major language now. Your users will thank you and it doesn't hurt you to support interfaces in as many languages as possible (obviously there are business and legal reasons to officially support only certain countries).

But it can be even easier. Why reinvent the wheel every time? All sites have common words like 'help', 'print', 'next' and 'previous'. Let's copy the Micro Formats people and have Micro Localizations. Peter showed a very basic demo: http://peter.beachheadonline.com/microlocalizations/index.php5. The page can load an xml file of either English or French words for common words, then uses the DOM to access and display them.

If Micro Localizations takes off there could be a central repository of language xml files, people could just refer to the central server to get what they need. Also, open source projects like OpenOffice and Mozilla already have huge localization teams, should reuse their work.

Jesus asked, could use have a user localization file just like user css? Several people considered whether there would be implications for accessibility. Maybe dyslexic people could use these files to make websites easier to read?

There was some discussion about other elements of localization that an xml file of common words wouldn't fix. For instance, typing many Asian languages is difficult, so many sites give lots of links instead (e.g. Google uses their Zeitgeist system to give the top 100 searches for that day). How do you support this?

• There was another presentation at DemoCamp to get collaborative translation of blog entries, etc. (Peter: Yes, I talked to Brian of Worldwide Lexicon people on Sunday). Also, open-source/php web sites usually just have a simple array text file mapping 'ENGLISHSTRING' => 'localized phrase'. We (Peter: Philip Jain?) implemented something like that in our international family networking site http://degree7.com. We used a templateized {@LANGSTR(ENGLISH STRING)@} function to basically map to a localized phrase if found, otherwise i defaults to the 'ENGLISH STRING'. Pretty easy to do, and we can support both unicode (all Indian languages,etc) also. Looks really good in many languages with very litte effort. If someone needs help send a email to info@degree7.com and can connect! In particular we are interested in getting help in Chinese, Korean, etc. languages.

• Another area for Web2.0 sites is getting the user to select MULTIPLE language content they are interested in. Veoh has a good user language selector, that we were thinking of emulating. This way we can show content in the preferred user languages.