QA mismatching segments


I have a tab-delimited bitext and I am running a checklist. The QA is flagging some segments where the target is not translated the way the rule says it should be be. What puzzles me is that in my translation, and in the bitext, those segments are correct. This happens in four or five instances in the file.

Here is an example:

My rule says the following, wth RegEx and PowerSearch activated:
“^2018 Expenditure$”

-"^2018 Gastos$"

The checklist flags three instances of

Source: 2018 Expenditure
Target: 2020 Estimación

BUT all three instances of “2018 Expenditure” are correctly paired with “2018 Gastos” in my tab-delimited text file. It is not a case of bad alignment, I have checked and the alignment is correct. I have copied my file to Excel and it also shows correctly paired. I have converted the Excel file back to tab-delimited and it still looks properly aligned, but when I run the checklist again it still says that there are three instances of “2018 Expenditure” wrongly paired with 2020 Estimación.

What could be going on here?

[Edited to add: this is Xbench 2.9]

I cannot seem to able to reproduce your issue. I tried both with Xbench 2.9 and Xbench 3.0.

Would it be possible for you to send a sample file with which we can reproduce the issue to support?