Sorry Sebastien Desautel  - I inadvertently deleted your answer, apologies! Here it is again:

"Hello Jean-François,

I worked on a similar project: the flow of the sdlxliff files in Trados Team is comparable to any other files. You can upload them in a new project and let them go through a standard workflow. TT detects the language direction of each file and will only prepare the file for the dedicated direction. You just have to take care that all language directions are available in the project template you are using, otherwise some sdlxliff files will be set as reference.

For the delivery, the translated files can be zipped and uploaded to WorldServer (I believe .zip extension is fine). That's at least what worked on the WS server we worked on. WS can find the internal path to upload the files through their metadata. So you don't have to create a specific return package.
As far as I can remember, you can also zip files from different projects together and upload them at once.

Kind regards

Sébastien"

Hello Sebastien Desautel ,

Thanks a lot for your input! I also found out that the files seem to be processed very straight forward, and that the delivery packages are nothing else than the zipped .sdlxliff files.

I have another question if you don't mind (and if Daniel Brockmann  doesn't delete your reply...  ) : The source files are already pre-translated and populated, however the source file packages also contain .tmx files and some reporting data. How flexible is TT in case we need to re-apply MT, TM, and re-analyze using a different fuzzy matches breakdown category?

Kind regards,

JF

Hello Jean-François,

TT will recognize the 100% matches (or better) as translated. For the untranslated segments, you can do whatever you want, like for a standard file: use your own settings (TM and/or MT) to pretranslate them, and yes this can be done with other fuzzy match categories. 

One thing you won't be able to change is the segmentation.

If you would like to leverage the segments exported as tmx, you will have to import them in the TM used for the analysis.

Use "Always overwrite existing translation" or "Overwrite existing translation if better match found." if the sdlxliff files are pretranslated with fuzzy matches and you want to prioritize your assets over WS pretranslation. 

WS pretranslation below 100% won't be taken into account in the analysis at all. The segments will be simply pretranslated and without any match in your TT TM, these segments will be counted as "new match" in the analysis.

Kind regards

Sébastien

Thanks again, Sébastien!

It's all clear and it makes sense. Can I ask for one more clarification if you don't mind? From what I understand, the .SDLXLIF files won't contain any fuzzy matches. So it means that if we reanalyze the files with the TMX included, we could (and should) leverage more fuzzy matches for the translators, correct? So the process would be to import the .SDLXIFF and the TMX, reanalyze the files and generate new reports?

Kind regards,

JF

Hello Jean-François,

yes, this is correct. and yes: 1. import the sdlxliff in a new project, 2. import the TMX in TMs (that are already set up in the translation engine to make this easy) and 3. start the project/workflow to get the analysis from TE.

The analysis algorithms are not exactly the same in TE and WS so you must expect some differences between the respective reports when only the TMX are leveraged.

Kind regards

Sébastien

Hello Jean-François,

yes, this is correct. and yes: 1. import the sdlxliff in a new project, 2. import the TMX in TMs (that are already set up in the translation engine to make this easy) and 3. start the project/workflow to get the analysis from TE.

The analysis algorithms are not exactly the same in TE and WS so you must expect some differences between the respective reports when only the TMX are leveraged.

Kind regards

Sébastien