The actual version of OmegaT includes an access to MyMemory as a machine translation provider : the result appears in the "machine translation" window. Not only this is not correct but this has the following inconveniences:

• It can retreive only one result (or two, since there are two providers, but not necessairly the two best ones);
• You see only the translation, not the source. In case this was a fuzzy match (less than 100 %) you do not see the score, so you have no way to evaluate the quality of the result

The first patch (M-1) introduces a modified version of the plugin to use it as a translation memory provider, rather than machine translation provider. Results of MyMemory are simply added in the matches list, then OmegaT applies the same rules as for results coming from TMX files: each result (not only the best scored one) is submitted to the fuzzy matches search thread, integrated in the correct location depending on its score, or not if it is too low. Note that at this step, we use the scores provided by OmegaT (all the three existing ones), not the score provided by MyMemory : as a consequence, results from MyMemory will have exactly the same aspect as other ones, only the "file name" is used to make the distinction.

Compare in the following screenshot:

In this sample, the left part contains machine translation, while right part contains matches: you can see that matches also contain source text (includiing differences), and do not necessarily only contain 100% matches.

We provide it as an example about the new "translation memory provider" interface. But we do not claim that it is complete. We did not make the distinction between MT and non-MT results, that could be another step. Also it would be possible to implement searchable interface (for inclusion in the searches window) but probably not the browsable (for auto-population), unless you know a way th retreive the entire contents from MyMemory...