An article for translators: Q and A with Tim Foster of Sun Microsystems

By Corinne McKay

Freelance Writer and Translator
ATA-Certified for French to English Translation

As reported in Open Source Update issue #8, Sun Microsystems recently released the first two applications in its Open Language Tools project. Tim Foster is an engineer working for Sun Microsystems, Inc. in the Globalization group, currently working on translation tools and technologies. He also is the project lead of the Open Language Tools, and maintains a blog about Translation, Language & Tools (and occasional off-topic items).

Open Source Update: When did the idea for the open language tools project get started? Were you involved with it at that point?

Tim Foster: Sun has been working on translation tools for quite a while : back in 1999 we started working with a 3rd party to customise their translation memory system for use internally. Over the course of the next two years, we started running into a few limitations with this system, so we set out to write our own system from scratch, based around more standard formats and implementation techniques. By mid-2002, we had the new system in production and had replaced the previous one. In October that year, we started thinking that the system could be of use to other people, particularly open source translation communities. I've been working on the translation tools at Sun since 1999 and was heavily involved in the re-write of the translation system.

TF: Well, it was pretty straightforward reasoning : 1. Sun isn't a translation tools provider 2. Sun works with a lot of translation communities, who may be able to use these tools to the benefit of everyone We went into this in a bit more detail in our initial announcement.

OSU: What was involved in bringing the tools from their in-house use form to the form in which they were released as open source?

TF: Well, it was largely a question of workspace layout and restructuring, moving code from java packages under the "com.sun.transtech" namespace to the "org.jvnet.olt" namespace. Internally, we had a single workspace which contained the XLIFF Filters, XLIFF Editor, utilities and translation memory system code. We needed to separate out dependencies so that the internal workspace could now reference the Open Language Tools workspace during it's build - we're now using exactly the same code internally as is in the open source workspace.

TF: So far, I think it's been very good, we've had a lot of hits on our project web page, and shipped around 3Gb of downloads in the week that the binaries were put on the project home page. We made the front page of java.net this week, so I'm really happy with the response so far : let's hope we can keep the momentum going - we'd be delighted to get new volunteers for the project and have a few suggestions on how people can help.

OSU: Can the editor use translation memories created with other tools? If so, what needs to be done to convert the TMs to a usable format?

TF: Historically, because the tools had only really been used in-house at Sun, we didn't put much effort into cross-compatibility with other TM systems. For example, in the first week of our release, users discovered some bugs with our XLIFF implementation, only found because up to that point, the only XLIFF files the system had needed to work with, were ones that were generated by itself. We have fixes for this problem in CVS I believe, and should be in the upcoming 1.1 release. Now that we're in the open, cross-compatibility is something that's becoming much more important - and we're committed to improving the quality of the tools in this respect. Right now, the editor uses it's own XML-based format for "mini-TMs", and it's possible to write XSLT scripts to transform TMX files into that format, it shouldn't be too complex to import other TM formats into the editor's mini-TM, but we haven't got this implemented yet : feel free to vote on Issue 12 and supply sample files if you can, which would really help us out !

OSU: What advice would you offer to other for-profit companies that are interested in open-sourcing their in-house software?

TF: I think the best advice I could give for companies in this situation, is to read a book called Innovation Happens Elsewhere, by Ron Goldman and Richard Gabriel. We used this as our main reference during the course of the tools open sourcing project. I just can't recommend this book highly enough ! There was an article on the Sun website about the book and its authors, which includes a sample chapter of the book. In particular, it's worth reading the section "Common Open Source Myths, Misconceptions and Questions".

This article may be freely reproduced or redistributed
for non-commercial use with attribution to the author
Copyright 2005 by Corinne McKay