Start a new topic

Unexpected behaviour with segment state (Auto-propagate forward only)

Segment state „Translated“

Filter view: Repeated and propagated segments

Segments:

Total                         Gesamt

Total                         Gesamt

Total                         Gesamt

Total                         Gesamt


Now I change the first one:


Total                         Summen

Total                         Gesamt

Total                         Gesamt

Total                         Gesamt


I see that the others don't change (should they? it would be nice to configure this behaviour), but in the next segment, I can enter the context match by pressing Ctrl-Option-4.


Total                         Summen

Total                         Summen

Total                         Gesamt

Total                         Gesamt


Now I am in segment 3, but I see that I made an error. So I use the "Move to the previous segment§ command and – surprise, surprise – the segments look like this:


Total                         Gesamt

Total                         Gesamt

Total                         Gesamt

Total                         Gesamt


Notabene: "Gesamt" is not part in any TM, "summe" was and is the exact/context match, and it is sufficient to go back into the 2nd segment to switch back also the 1st. 


This behaviour is a lttile bit irritating, no matter whether I have "Auto-propagate forward only" activated or not (indeed I do). With "Auto-propagate forward only" deactivated this does not happen.


It looks like sorting for repeated segments in the filter might be reversed for the next build.  

Just as an illustration:

image


A bit of "Hide and seek" (here still rather easy) for anybody with the setting "Auto-propagate forward only" (in my humble opinion only a kind of emergency mode if things go wrong, but at the same time the default setting in CT, if I'm not mistaken).

 


> I don't think it really matters in what order you filter the repeated segments since they are taken out of their contexts.

Yes, that's true, but only for auto-propagate in both directions (although in combination with "Display current segment context" this argument loses a bit of validity). And it is only about sorting the filtered results in their natural order (assuming this is not easy to code, but seeing the necessity).

 

CafeTran already performs the identical segments grouping in the filter so the whole filter view can never be sorted in the order of appearance of the filtered segments. Otherwise, it would end up with identical segments being shown separated between other identical segments. I don't think it really matters in what order you filter the repeated segments since they are taken out of their contexts. Only the identical subgroups might get additionally sorted for convenience in some future update. But that's easier said than done in terms of the filter speed with big projects containing several glued documents with lots of repetitions.

My humble opinion is to expect that repeated and/or repeated segments are sorted – also in this filter – by their appearance in the document, not by random.

 

Depending on your settings, this might mean that segments are treated and filled differently (eventually arbitrarily) depending on filtered or unfiltered view. Really?


Else:

What then about a filter "Sort by appearance" besides "Sort alphabetically" and "Sort by length" (though this appears a bit strange)?


 

> Filter view: Repeated and propagated segments


The sorting is as expected in the name of the filter.

Perhaps a normal user does not assume that the order of the filtered segments is not the same as their real order. Let's call it "unexpected behaviour".


What about to avoid this "faulty" sorting in the filter (IIRM I already noted this in another thread)?


The other thing is that no segment with the false "Gesamt" is being confirmed above, so I wonder why segments are being set back to it. It simply should not be the case, should it?

Your confusion comes from the fact that you use "Auto-propagate forward only" option with a filter where the order of the filtered segments may not the same as their real order.

Login to post a comment