It sounds to me like a bug in that software, something like a bad
regular expression that is catching too much text. FYI, I'm guessing the
"functional characters" are something like the "[=" and "=]" that seem
to surround these citations.
FWIW, as well, this sounds like a needlessly complicated system. Why not
just use a BibTeX editor, or have this program export a BibTeX file, and
use that for the references? If you ultimately need a file that has just
those references, then you can use the aux2bib program to make one.
Richard
Thank you for your reply, Richard. But switching to a different
bibliography software isn't really an option since I would have to
replace over 340 references in the text.
I'm now closer to understanding the problem. When I viewed the source
code in LyX (the LaTeX code) I couldn't see that something in the way
the files were coded had changed. I used Windows Editor to open the two
.lyx files, the one before and the one after the compilation, and I
noticed that the umlauts (I'm writing my paper in German) looked normal
in the original file (e.g. "ö" and "ü") but in the new file they had
been transformed into codes (e.g. "ö" and "ü"). Now LyX didn't seem to
have any problems with the coded umlauts although they were different
from the original file. But the ordinary looking ones that appeared in
some of the references generated by Bibliographix caused the
corresponding text passages to disappear in LyX.
I googled "ö" and found out that it's UTF-8 code. But I don't know what
to do with that information.
Obviously, I can use the replace-function of the Editor to transform the
umlauts into UTF-8 and I suspect that that's as close to a solution as I
will get with this problem. But I wonder what caused it.
Peter