biber produces empty .bbl file.

[Haines Brown]

In ting to customize biblatex-ieee, I managed to break the system and so
removed TexLive 2014 and installed TexLive 2015. Otherwise everything is
the same. After installation I could run pdflatex on sample2e.tex
without trouble.

I logged out to get the new path working:

$ echo $PATH
...:/usr/local/texlive/2015/bin/i386-linux:...

I'm using a simple biblatex.tex file with a citation:

\documentclass[12pt]{article}
\usepackage[utf8]{inputenc} %
\usepackage[T1]{fontenc} %
\usepackage[authordate,backend=biber]{biblatex-chicago}
\addbibresource{/home/haines/jabref/nat-sci.bib}
\begin{document}

citation: (\cite{Barrett2007}).

\printbibliography
\end{document}

I run

$ pdflatex biblatex.tex

The output includes these lines that I expect:
LaTeX Warning: Citation 'Barrett2007' on page 1 undefined on input \
line 27.
LaTeX Warning: Empty bibliography on input line 33.
LaTeX Warning: There were undefined references.

It produces the .pdf and .bcf files

I next do:

$ biber biblatex.bcf
INFO - This is Biber 2.1
INFO - Logfile is 'biblatex.blg'
INFO - Reading 'biblatex.bcf'
INFO - Found 1 citekeys in bib section 0
INFO - Processing section 0
INFO - Looking for bibtex format file \
'/home/haines/jabref/nat-sci.bib' for section 0
INFO - Decoding LaTeX character macros into UTF-8

this produces .bbl and .blg files. However, the .bbl file is empty.

the .blg file says

[0] Config.pm:324> INFO - This is Biber 2.1
[0] Config.pm:327> INFO - Logfile is 'biblatex.blg'
[38] biber:276> INFO - === Thu Oct 1, 2015, 19:37:03
[38] Biber.pm:333> INFO - Reading 'biblatex.bcf'
[138] Biber.pm:645> INFO - Found 1 citekeys in bib section 0
[149] Biber.pm:2977> INFO - Processing section 0
[157] Biber.pm:3115> INFO - Looking for bibtex format file \
'/home/haines/jabref/nat-sci.bib' for section 0
[1282] bibtex.pm:1021> INFO - Decoding LaTeX character macros \
into UTF-8

Because the .bbl file is empty, when I run pdflatex again, I get:

LaTeX Warning: Citation 'Barrett2007' on page 1 undefined on \
input line 27.
LaTeX Warning: Empty bibliography on input line 33.
LaTeX Warning: There were undefined references.

[Herbert Schulz]

Howdy,

I'm Wondering if Barrett2007 is actually defined in
/home/haines/jabref/nat-sci.bib? Have you tried to copy that bib file
into the same directory as your source file 9and change the reference
to it of course)?

Good Luck,
Herb Schulz

[Johannes Böttcher]

On 10/02/2015 02:10 AM, Haines Brown wrote:
> the .blg file says
>
> [0] Config.pm:324> INFO - This is Biber 2.1
> [0] Config.pm:327> INFO - Logfile is 'biblatex.blg'
> [38] biber:276> INFO - === Thu Oct 1, 2015, 19:37:03
> [38] Biber.pm:333> INFO - Reading 'biblatex.bcf'
> [138] Biber.pm:645> INFO - Found 1 citekeys in bib section 0
> [149] Biber.pm:2977> INFO - Processing section 0
> [157] Biber.pm:3115> INFO - Looking for bibtex format file \
> '/home/haines/jabref/nat-sci.bib' for section 0
> [1282] bibtex.pm:1021> INFO - Decoding LaTeX character macros \
> into UTF-8

The `biber' run did not finish. Please run the following example *as is*
and report the result:

%\documentclass{article}
\usepackage[style=authoryear,backend=biber]{biblatex}
\addbibresource{biblatex-examples.bib}
\begin{document}
Zitiere Buch: \parencite{companion}\par
\printbibliography
\end{document}
%

[Haines Brown]

I don't understand. You say "as is", but without \documentclass{} it
won't compile.

! LaTeX Error: \usepackage before \documentclass.

I removed the initial "%" and now latex runs properly. So does biber,
and the resulting PDF is OK.

The problem not that TL 2015 biber works on bib files in general, but
not for two of my seven bib databases. The problem exists within these
two databases.

So I ran this same TL 2015 biber on an old backup of one of these two
that had run successfully with TL 2014. It exits rather than
complete. So it seems that biber is exiting because of something in that
database that TL 2014 could handle. This suggests either a bug in TL2015
biber (can't cope with .bib files over 5 Mb ?) or perhaps its standards
are more stringent than TL2014 biber.

[Johannes Böttcher]

On 10/03/2015 07:20 PM, Haines Brown wrote:
> I don't understand. You say "as is", but without \documentclass{} it
> won't compile.

Sorry, this was my mistake, i forgot to hit enter after the comment
which was supposed to be some sort of start sign. I did not check :-\

>
> The problem not that TL 2015 biber works on bib files in general, but
> not for two of my seven bib databases. The problem exists within these
> two databases.

That is what i wanted to know, biber is ok, but the databases are
causing the trouble. I bet one entry is causing the crash. Can you try
to find the entry using the same algorithm as in preparing a MWE?

>
> So I ran this same TL 2015 biber on an old backup of one of these two
> that had run successfully with TL 2014. It exits rather than
> complete. So it seems that biber is exiting because of something in that
> database that TL 2014 could handle. This suggests either a bug in TL2015
> biber (can't cope with .bib files over 5 Mb ?) or perhaps its standards
> are more stringent than TL2014 biber.
>

I don't think that the file size is an issue, but there indeed have been
changes with 2015. But that should be noticable at a later point in the
compilation chain.

[Haines Brown]

Johannes Böttcher <johannes...@gmail.com> writes:

> That is what i wanted to know, biber is ok, but the databases are
> causing the trouble. I bet one entry is causing the crash. Can you try
> to find the entry using the same algorithm as in preparing a MWE?

You loose me. A minimal working example of what? Was my test file not a
minimal working example? Assuming not, I tried this:

> \documentclass[12pt]{article}
> \usepackage[utf8]{inputenc} %

> \usepackage[backend=biber]{biblatex}%
> \addbibresource{/home/haines/jabref/soc-sci.bib} %
> \begin{document}
> Soc-Sci: \parencite{Adams2001}
> \printbibliography
> \end{document}

And biber still exits prematurely. On my system, the exit does not
indicate which line or field was causing it to do so.

As I write, I'm installing TL 2014 from cdrom disk (couldn't figure out
where to direct the installer to a compatible CTAN version). If for some
reason this also can't use the soc-sci.bib database, I'll report
back. It would mean the problem isn't with the TL 2015 biber, but
something really bad that somehow got into the database.

If contrary to my expectations, this is the case, then lacking a tool
that can identify the problem in the database (do a trace?) I'll have to
do the tedious job of rebuilding a new .bib file from pieces of the old
until I pin down the offending record.

[Johannes Böttcher]

On 10/03/2015 09:11 PM, Haines Brown wrote:
> Johannes Böttcher <johannes...@gmail.com> writes:
>
>> That is what i wanted to know, biber is ok, but the databases are
>> causing the trouble. I bet one entry is causing the crash. Can you try
>> to find the entry using the same algorithm as in preparing a MWE?
>
> You loose me. A minimal working example of what? Was my test file not a
> minimal working example? Assuming not, I tried this:
>

> As I write, I'm installing TL 2014 from cdrom disk (couldn't figure out
> where to direct the installer to a compatible CTAN version). If for some
> reason this also can't use the soc-sci.bib database, I'll report
> back. It would mean the problem isn't with the TL 2015 biber, but
> something really bad that somehow got into the database.
>
> If contrary to my expectations, this is the case, then lacking a tool
> that can identify the problem in the database (do a trace?) I'll have to
> do the tedious job of rebuilding a new .bib file from pieces of the old
> until I pin down the offending record.
>

I think there is indeed an offending entry in your database file, but we
don't have access to it. You mentioned it is 5 MB large, so you cannot
post it completely here. Rebuilding the database (or commenting out bits
and pieces in a strategic way) will lead to finding the cause, which is
indeed the same algorithm as in preparing an MWE.

PS: Tried emailing you, but wasn't successful.

[Haines Brown]

Johannes Böttcher <johannes...@gmail.com> writes:

> I think there is indeed an offending entry in your database file, but we
> don't have access to it. You mentioned it is 5 MB large, so you cannot
> post it completely here. Rebuilding the database (or commenting out bits
> and pieces in a strategic way) will lead to finding the cause, which is
> indeed the same algorithm as in preparing an MWE.

That's what I'll do. If I find a problematic field that passes TL 2014 biber
and fails TL 2015 biber, I'll report it back here.

[Peter Flynn]

On 03/10/15 20:11, Haines Brown wrote:
> Johannes Böttcher <johannes...@gmail.com> writes:
>
>> That is what i wanted to know, biber is ok, but the databases are
>> causing the trouble. I bet one entry is causing the crash. Can you try
>> to find the entry using the same algorithm as in preparing a MWE?
>
> You loose me. A minimal working example of what? Was my test file not a
> minimal working example?

No, it needs the bib data.

You said you have several .bib files. Run it with one of them only, then
again with the second one added, then add the third, etc until it fails.
They you know which .bib file contains the error.

Now edit the .bib file: find the half-way mark and delete everything
after that. Run LaTeX and see if it still causes an error. If so, the
error is in the top half; if not, in the bottom half.

Repeat, cutting it at the quarter or three-quarter mark respectively,
according to which half it failed on. Re-run. Repeat recursively,
cutting the file down progressively until you isolate the error.

This is hugely tedious, but guaranteed to find the problem.

///Peter

[Haines Brown]

This is more or less what I did, and found that I had tried to escape
"=". Apparently TL 2014 biber didn't object, but TL 2015 wouldn't
tolerate it.

[Axel Berger]

Peter Flynn wrote on Sun, 15-10-04 00:43:
>This is hugely tedious,

Does biber not offer verbose logging?