This article was kindly provided by Michael Beijer.
Yes, the terminology options in CafeTran are a little confusing, basically because there are various ways to achieve pretty much the same thing: create and use a termbase.
Basically:
• translation memories are stored in TMX files.
• termbases can take the form of:
1. tab-delimited text files (called 'Glossaries' in CafeTran jargon), or
2. TMX files (called 'Memories for Terms' in CafeTran jargon)
• you can also store resources (terms, longer phrases, or entire segments) in so-called external databases, but this is only for lookup; they are not connected to the auto-assembly system
They each have their benefits. Glossaries are good for if you wish to use synonyms (source-side or target-side synonyms), which is one of the reasons why I use them. However, there are other CafeTran users who swear by the Memories for Terms, for example. There is no one way to do things, they are all good in their own way.
In the latest version of CafeTran, Igor added a new category in the Project Manager, which he calls 'Phrases' (See the screenshot below), which CafeTran stores in TMX files. This basically means that Igor suggest you store longer phrases in TMX files. However, I disagree, and actually just put all my terms (whether single terms or longer phrases) in a .txt Glossary (the section called 'Terms' in the screenshot). I use metadata (the fields that must be present in the tab-delimited .txt Glossaries, which can be defined at: Edit > Options > Database) to make sense of all my terminology in a single .txt Glossary.
I suggest:
1. using a .txt Glossary for all terms (short or long) using the section called 'Terms',
2. creating a project translation memory using the section called 'Segments', and
3. using a background (or big mama/papa; or whatever they call them) translation memory (if needed), from the section titled 'Phrases'.