I have long been curious about OmegaT, mostly for 2 reasons:

1. OmegaT is the last free CAT on the market still being developped (that I know of)
2. OmegaT is written in Java, and therefore relatively platform independent.

• Installation:

  Assuming you already have Java J2RE and OpenOffice.org installed, installing OmegaT itself is a breeze. You just unzip it in a folder, and that's it. However, installing Java and OOo will be a bit more demanding, and you have to download 'em too, probably one of the reasons why few people currently use OmegaT.

  On Linux, installation/running is a tad more difficult -requires writing a small shell file- but, hey, what's new? (The explanation on the manual is nearly good enough for a newbie to figure it out with minimal tweaking!)

• Translation process:

  When you run OmegaT, it shows up in 2 separate windows, one for the translation, the other one for matches and glossary. First, you have to create a project, which is pretty easy. Click on “Files/Create New project”, select the folders (the default folders will be best usually).

  Then, you have to copy your source files in the source folder of the project using your file explorer. If you have TMs (in TMX 1 format), copy them in the TM folder. Same procedure for glossaries (in tab delimited format). Once this is done, you are ready to start the translation: Go to File/Open, and select the project file (which has been created for you in the root folder of the project).

  Your file will appear in the main window. Press Enter to start the translation. When you are done with the segment, just press Enter. To return to the previous segment, press Ctrl+P. You simply press Enter to move on. Real easy.

• File formats.

  OmegaT handles text files, HTML files and any file that can be opened in OpenOffice.org. That includes RTF, and most importantly Word/Excel documents, thanks to a series of powerful filters. However, filters are not the real thing and some intricate formatting could be messed up. Still, it remains OK for the very large majority of the jobs.

• Segmentation

  This is one of the most important differences between OmegaT and other CATs. OmegaT will use paragraphs as segments – instead of sentences. I will discuss the advantages and downsides of this approach later.

• Fast

  Changing from one segment to the next is near instant, at least on average files. (Haven't tried anything longer then a few thousand words).
  
  According to the manual however, OmegaT does get much slower when several large files (containing many segments) are added to the “source folder”. The work around is however not too difficult. See the ASAD manual for more data if you come across this problem
  
  I have experienced no noticeable slow down with 5000 words, which is a fair size for an individual file.

• Paragraph segmentation

  This is very interesting for literary, marketing and legal documents. Indeed, to produce a good translation on a marketing document, it is often necessary to read a full paragraph and rephrase it, rather then going sentence by sentence. Not all documents can be translated sentence by sentence, and this approach may result in a poor translation. Each sentence may have been translated correctly, but the style is lost. It doesn't flow. On a marketing document, translation often requires one to rephrase a whole paragraph and combine sentences*.

• Platform independent

  Can be gotten to run on any platform...BUT it requires Java J2RE, which is seldom present on Windows machines, and to do any real work, OOo is a must – free too, but 60 some Mb to download.

• Simplicity of operation

  Once you have learned to press Enter to change segment, and Ctrl+P to go back to the previous one, that's it. There are nearly no bells and whistle, and nothing to distract your attention, so you simply focus on the translation.

• File structure.
  
  A nice thing to note (especially for website localization) is that OmegaT preserves the structure, including nested subdirectories and all non translatable files, in the final output. This makes it a tool of choice for website translation. (The OmegaT tags however need to be learned to do a good job, because you can't see the actual tags)

• Search function
  
  The search function of OmegaT is quite nice, with support for a couple wildcards, * and ?. You can search the TM as well, and you can also use a keyword search (a bit like an internet search engine: search for “game” and “bob”, and it will bring up all segments which contain those 2 words).
  
  Double-click on the segment you are interested in and you will be brought to the segment itself.

• Preview
  
  Preview is not the right term, but that's close enough. You can compile the files at any time during the translation, and it is reasonably fast. This means you can always see were you stand in the final file, a most accurate preview. This is very useful when working with HTML files, as you can see what it looks like, and what goes where.

• Price.

  What's cheaper then free?

• Open Source/GNU
  
  Open source means that you can access the source (Duh!) so if you are able to program in Java, you can customize it as you like/need, and integrate it with other java programs. While this is clearly not for the average user, it could prove useful later. If you spend your days on a computer, soon or later, you will dig a bit of programming, and when you do, Open Source is a dream come true.

• Glossary

  The glossary function is very clumsy. You have to add terms by opening the glossary in a text editor. OmegaT needs to be restarted if you want it to recognize new terms from the glossaries, and when it recognizes them, you still have to type the translation yourself.

• Display.

  You have two windows standing side by side, and they just don't integrate. The GUI looks poor and that's a bit of a distraction. Font settings are skimpy.

• Paragraph segmentation.

  It also means that there are much less matches than with a regular CAT. For instance, the following paragraph:
  “Go to the File menu, and select Save As. Enter a name for your file. Select a format and click on Save. ...”.
  With a typical CAT, these 3 sentences could be recognized and translated automatically. With OmegaT, the whole paragraph has to match, and it seldom does. As a result, TM brings much less matches then on regular CAT tools.

  Another downside of paragraph segmentation is that there is no real compatibility with other CAT products. You have TMX 1.1 TMs, but they bring no matches when used on other CATs. However, the search function can ease the problem a bit, as you can query the TM directly.

• Search function issues
  
  The segment number given is wrong (as per my trial).
  The search window remains in focus after you double click on a result, so you may not see the segment and it feels like nothing happened.
  Once you got to the segment you searched, there is no easy way to get back to the segment you were previously working on.

• Spellchecking/other functions
  
  “Simple” features like spellchecking require major tweaking to install. You can tweak and tweak and even reprogram the whole thing -it's GNU- but obviously most users are not computer litterate enough to do so, so the have to do without, for now.

• Tag handling.

  You can see where the tags are and move them quite easily, but you don't know what they are. That means you have to keep a copy of the document open to check what the tags are. That said, OmegaT can compile the files at any time, so you can easily get a preview.

  Because it handles the translation in a simple text editor, OmegaT turns simple word/rtf files into tagged documents, so you need to know how to handle tagged files, and you need to check the tags integrity, even for simple *.rtf files. Further, if you want to add formatting, you will have to do it outside (in OOo for instance), because it can't be done in OmegaT.

• Lack of support for other CAT formats

  Apart from a serious amount of tweaking, you can not use OmegaT for Wordfast/Trados files. Since OmegaT is in a minority position on the market, compatibility is important. A translator doesn't translate alone, but integrates within a team, (be it his own team or that of his customer) and must therefore be able to exchange easily his work files with others. That said, if you are ready for the tweaking, it can be done, but not within OmegaT itself.