kjn: (Default)
[personal profile] kjn
I've done some work that is out in the wild, but mostly my text ends up within much larger amounts of text (like the help in Mac OS X), or they've been through at least one editing pass after I hand off the translation.

However, last Saturday, the newest and greatest version of iCalamus, a publishing program, for Mac OS X, was released. And the translation and localisation to Swedish there is mine, all mine (well, except for my excellent reviewer/proofreader, but I entered the edits myself).


The other localisations of of iCalamus are done using iLocalize (for the software) and TextWrangler (for the help and web pages). The software is standard OS X bundles, and the web pages are simple HTML. The help files are written in an old, obscure format called UDO, which are transformed into the desired formats later on.

iLocalize is an excellent localisation tool, but it's not a translation tool other than in the most basic sense. Its strengths are in integrating translations into the main package (useful for developers) and in doing fast controls and lookups of the UI. Its translation memory handling (glossaries) is awkward and clumsy compared to purpose-built CAT tools, and the editing environment could be a lot better. TextWrangler is a good, basic text editor with excellent support for character tables, but it's not a translation tool at all.

I decided I could do better—which I did. The software bundle was converted via iLocalize (nibs and some other special files) and AppleGlot (strings files) so I could translate the software in SDL TagEditor. There were lots of nooks and crannies and special cases to both iCalamus, iLocalize and AppleGlot, and it probably took two days to actually find and handle all of them, but once it was all translated I just had to export a translated file from TagEditor, import it to iLocalize, and the software was almost fully localised.

The help files were next. UDO is, at its core, tagged text, but it's not XML or SGML,
instead more akin to GNU Texinfo. TagEditor has a regexp-based filter for formatted text. That's the good news. The bad news is that said filter engine was not documented at all, it was almost impossible to edit the filter rules, you could only ever have one filter at all saved at the same time, the filter handling was extremely clumsy, and all the error messages were the computer science equivalent of "Thou shalt not seethe a kid in his mother's milk".

But, in the end, I got a somewhat workable version. Not perfect, but workable. TagEditor even managed to output valid translated UDO files back, except where I had messed things up. The web pages were a breeze after that. And I must say that working with the software and its documentation using the same tools and translation memories was a godsend. I probably saved back all the time lost in building my custom workflows just in the unified editing and memory environment.


So that is how I spent much of December and January.

Me

kjn

Links

Tags