just heard about an interesting open source TM server thingee, made by the Swordfish guy:
Rodolfo M. Raya firstname.lastname@example.org via groups.io
6:48 PM (2 hours ago)
TMEngine is an open source Translation Memory Engine written in Java, based on the translation memory library used by Swordfish III, Fluenta and RemoteTM.
TMEngine can be used in two ways:
As a standalone TM server via its REST API.
As an embedded library that manages translation memories in a Java application.
With a TMEngine server you can share Translation Memory data in a local network and over the Internet.
The standalone server runs on these platforms:
Microsoft Windows 8, 1.1 and 10
macOS 10.13, 10.14 and 10.15
Linux (any version capable of running Java 11)
TMEngine source code is available on GitHub, at https://github.com/rmraya/TMEngine
TMEngine and OpenXLIFF Filters are the foundations for an open source TMS (Translation Memory Server) that is currently in development.
Send feedback, report errors and request new features to email@example.com
That would be very interesting.
What would it add ...
I was hoping: the ability to connect a very large TM to a project and use it live.
For example, in memoQ, you can use the Wordfast plugin, and so just dump literally millions of TUs in a Wordfast TM (= a simple tab del!), and you get matches in memoQ while translating, with difference highlighting. Very cool.
It is still rather difficult to work with very large TMs in Cafetran. It can't handle opening large TMXs (no matter how much RAM you have), and if you opt to use Total Recall, you still run into problems when trying to quickly start a new project and immediately access very large TMs.
Yeah, I was aware of your approach. But, with at least 5 MT systems for NL > EN active, is it still worth your while?
Ha ha, probably not ;-)