Start a new topic

Match repair: duplication

Since a couple of days I get a duplication when match repair is triggered:


image


A=A and B=B are in my glossary, I get this A-A result:


image


How come?


I've added two regular expressions to the non-translatables glossary for this client:


image


Both regular expressions are functional:


image

and:


image

Match repair still places the R behind the (A)

On third thought: that's not exactly true, I now get a match repair (in CafeTran jargon: auto-assembling fuzzy match correction) but with double quotes:


((R))

Another example:


image


and:


image


Same here. Match repair does no magic any more... 

Yes, there is a systematic duplication active now. Currently translating a manual where the company name is duplicated every time when match repair is triggered.

To be more precise: Müller Systems GmbH is constantly being "repaired" to Müller Müller Systems GmbH.

I wouldn't be surprised if that's caused by tab dels because they don't allow for fuzziness. It's all or nothing.


H.

I think that it is there since the introduction of Automatic fragments adjustment.

Thanks for reporting this issue. I will look into as soon as possible - probably next week.


2 people like this

IK: I think that it is there since the introduction of Automatic fragments adjustment


Long before, and I'm sure I mentioned it. And I think the problem disappeared after I switched to TMX only, again, long before Automatic fragments adjustment.


H.

Match repair still places the R behind the (A)


I don't see any duplication issue here: BG36 (A) R. CafeTran appends R to the fuzzy match because:


1. It is at the end of the source segment.

2. You have R entry in your resource.


Another example:


In the second example, there is a case difference between source fuzzy match and current source segment (at the end of both). However, CT still appends the last target entry despite it is already present at the end of the TM fuzzy target segment (no case difference here). The check will be added to prevent this sort of duplication at the start/end of TM fuzzy target match. Thanks! 



 

@Hans: 


re:

To be more precise: Müller Systems GmbH is constantly being "repaired" to Müller Müller Systems GmbH.

 … didn't you post something recently about a regex non-translatable, with that term ("Müller Systems GmbH")? Could this be causing the trouble here?

Hm, I am unsure if this issue has been really resolved. Following scenario:


image


"mesure" is in the TB (and nowhere else) as "Messung". This is what the target segment automatically looks like:


image


Why this duplication? Without the duplication everything would be perfect.

Please check it with the latest build 2017091301 of the update 21.


1 person likes this
Login to post a comment