Just some reference: \textcite{myfoo}
...
\usepackage[backend=bibtex]{biblatex}
...
2) Run it with texi2dvi -p mwe.tex
=> the script never stops
Thanks for the report. The issue is specific to biblatex, so I'll cc
Philip for advice. (Philip, the original mail is at
http://lists.gnu.org/archive/html/bug-texinfo/2014-04/msg00030.html.
Although the given example uses a koma-script class, I don't believe the
problem is specific to that.)
Back in August, I was advised
(http://lists.gnu.org/archive/html/bug-texinfo/2013-08/msg00044.html) to
check for
biblatex.*(re)run
in the .log file, and if it was there, take it at its word.
Now, I can easily omit that check. Then texi2dvi would fall back to
checking to see if the xref files it knows about have changed (and if
they haven't changed, then give up). The question is, is that
sufficient for biblatex? In the present case, it would be checking
these three files (I named the test file bibunref.tex):
bib.bib
bibunref-blx.bib
bibunref.aux
I see there is also bibunref.run.xml to perhaps be checked?
And I don't know if other files get created in more complicated
situations.
Anyway, as far as the infinite loop goes: that is clearly bad. You can
specify --max-iterations=3 to make it stop after 3 cycles. The default
before was infinite; I've just uploaded a new version (to ftp.gnu.org,
if you want to try it) making the default be 7. If things haven't
converged after that many runs, I sincerely doubt they ever will, as a
practical matter.
Thanks,
Karl
