Lots of interesting news today. Yesterday, Yves Savourel, the developer of the Okapi tool set, sent me a note saying this:

Just a word to let you know that a new set of Java tools has been released today. The "old" .NET/Windows-based Okapi Framework has been completely redesigned and rewritten in Java. This allows the components to run on Windows, Linux and Macintosh. Moving to Java also made it possible for a few developers to join the new project and bring innovative ideas and more experience.

In this first milestone we have Rainbow v6, Ratel (an SRX editor), and a basic library. But the most interesting part is probably the common resource model all the components use, as well as the pipeline mechanism. This will be the backbone of the next releases. It will allow the users to really customize the tasks they need to run, and to create new tasks as well.

You can get the new packages from the Google Code site and you can get more information (including the beginning of a Developer's Guide) from the main site. We still have many things to develop, but hopefully this first milestone will give you a glimpse of what is coming.

All right, you might say, so what exactly is he talking about here? Well, Yves is the largely unsung hero who for years now has worked and released an open-source tool suite that deals with the often overlooked aspects of TEnT management, including conversions, maintaining TMs, and quality assurance. His tools include Olifant, the most advanced translation memory editor on the market -- this tool has not been ported to the Java platform, the new SRX editor that Yves mentions (SRX is the segmentation exchange standard that allows you to exchange the segmentation rules of your TEnT with another) - and, of course, the tool Rainbow. Rainbow is indeed an appropriate name for the tool: it seems to cover an amazingly wide spectrum of desirable features, including:

·         Encoding, line-break, URI, RTF, Byte-Order-Mark conversions

·         XSL transformation

Or for the less technical among us

·         Translation package creation

·         ID-based alignment

·         Translation comparison

·         TTX checker

Let's just look at the last four features.

The translation package creation allows you to create an XLIFF package or a package for OmegaT, Trados RTF, or Wordfast Classic from source documents that can be translated in the respective tools; the ID-based alignment allows you to align files with strings that have some kind of an ID (for instance Java .properties files) and comes with a full-fledged Alignment Verification editor; the translation comparison feature allows you to compare two versions of translated files to generate a score and an HTML or TMX report with the differences highlighted; and the TTX checker allows you to locate problem areas in Trados TTX files if they don't want to export properly. So, you can see, Rainbow is sort of a smorgasbord of things we often thought would be nice to have in commercial TEnTs.

I also really liked Ratel, the SRX editor. (I will need to ask Yves about that name. I'm sure there is some kind of clever explanation for it -- after all, "Okapi" relates to the "copacetic animals that are no strangers to innovation: they have blue tongues which they use to clean their ears. . . ." If that's not clever!) SRX is one of those standards that is only very slowly being adapted. The idea of SRX is great: The two large obstacles that TMX as a translation memory exchange format has to deal with are differences in segmentation and differences in codes within segments ("in-line codes"). SRX deals with the first issue by allowing tools to exchange an XML-based file that contains the necessary information about where a text is to be segmented and with what kind of exceptions. Unfortunately, the only tools that support SRX are MemoQ, Swordfish, Heartsome, SDLX, and XML-INTL.