Grouping translations across <trans-unit> elements

Hi,
I have a question about best practice of the new XLIFF 1.2 grouping of translation units.
In the XLIFF 1.2 draft it shows an example where the ending/beginning white space is not present in the source trans-unit but present in the translation of the merged trans-unit. How would the translator know that a whitespace should be added? Is there any best practice for this? If the whitespace between segments is in a skeleton file it will not end up at the correct place on back conversion and the merged source segments will be incorrect. 
A possible solution for this would be to add all whitespaces between translatable trans-units as separate trans-units with translate attribute set to "no" (or set to already translated) so white space information can be included in a merged segment.
What are your thoughts on this?
 
Regards,
Fredrik

________________________________
Fredrik Corneliusson
Software Engineer Language Technology
Lionbridge
F顤sta L幩ggatan 30
S-413 27 G飆eborg, Sweden
Phone +46 31 7756088
www.lionbridge.com <http://www.lionbridge.com/>