I find this function very of tremendous help. But, when yesterday I had to use it in the middle of a project, upon document reload all target segments were gone. I should say that none of these segments were in the TM (my choice), hence my doubt: to have the aleady translated segments reload untouched with the new document, should they have been in the TM?
This is what I have understood as well, an up-to-date Project TM must be attached for the replacement to work.
Maybe Igor can provide a confirmation of this feature's inner workings.
Anyway, I would always recommend backing up the project folder prior to conducting this operation.
Also, if you are working with a multi-document project, make sure you are replacing the current document.
Thank you Jean. As a matter of fact, I do many translations without sending segments to the TM when I deem them not meaningful enough for future jobs. My two gods are glossaries and fragments, and this only justifies using a CAT tool for me.
But, if sending segments to a TM is a pre-requisite for not losing them upon source documents replacement, and I hope Igor will shed some light on how this function actually works, I need to rethink the way I use CTE.
Source document replacement does not require any external translation memory. CafeTran treats the existing segments in the project as exact matches for the source segments of the replaced document. Of course, any changed source segments need to be re-translated manually. Moreover, the segments that you split or joined while translating the previous version of the document are not exact matches for the new version of the source document and are not be translated automatically.
Igor: Source document replacement does not require any external translation memory
Thank you for clarifying. Loosing all target segments occured to me upon replacing the source document with an only slightly modified file in, perhaps, one tenth of the whole text. I don't remember all details unfortunately, but in the next days I will try to do some tests on old projects to see whether I can replicate this behaviour.
If your old projects were created in another version of CafeTran as the one which does the source document replacement, the automatic insertion of exact matches may not work as expected. The filters for source file type (e.g Word) may be improved (changed) in next versions.