My glossary entries:
The result of auto-assembling (simply inserting from left to right, not looking for what is the longest possible match):
CT looks for the longest possible match in the linear fashion from left to right when assembling to target segment.
Okay guys, I've investigated this matter now. Long story short: with TMX files for terms you cannot let CafeTran insert the longest string either. It still works from LTR.
Auto-assembling result from a glossary
Please have a look at the glossary entries:
The auto-assembled translation:
Auto-assembling result from a TMX file
Via Memory > Import > Import entries from glossary I have imported the glossary entries to the ProjectTM. Then I have quit CafeTran and restarted it, this time using the ProjectTM only. Oh chips... I have forgotten to set it to TM4terms. :(. Okay, I'll quickly check again and report back if this changes this result:
Now I have renamed the ProjectTM to TM4Terms and opened it with the Termbase checkbox ticked:
This doesn't change the auto-assembling result:
The solution is quite simple (I had forgotten it -- again!): you shouldn't add glossary entries that can form the left-hand part of a longer entry (you can add glossary entries that can be the right-hand part of a longer glossary entry).
I will remove glossary entries with preposition + 'einen' + \t (and similar entries) from my glossary.
Just a note about the forum format, as I just wanted to reply to one of the posts in the middle, but can't. I think this really needs to be possible. Igor, if you are reading this, can you look into this?
Hans, as far as I understand it, whether the Termbase checkbox is ticked or not only governs whether terms can be saved to it, not much else.
As for the forum format, we can only wait and see if Freshdesk developers improve it in the future updates to the system, which is already a huge step compared to the Cafetranslators forum in terms of organization.