refdb-users

[Refdb-users] refdbib segfault

[Dylan O. <lo...@st...>]

Hi,

I started a document with refdbnd (docbook xml 4.1.2), inserted some
references to sources I added to my database (mysql), and have been
trying to compile for hours now without any luck. Here's what I'm
seeing:=20

make html                                =20
runbib -d rdbib_dao -S J.Biol.Chem. -t db31x cepprop.xml
/usr/bin/openjade:cepprop.xml:76:4:E: cannot find "cepprop.bib.xml";
tried "cepprop.bib.xml", "/usr/local/share/sgml/cepprop.bib.xml",
"/usr/local/lib/sgml/cepprop.bib.xml", "/usr/lib/sgml/cepprop.bib.xml"
/usr/bin/openjade:cepprop.xml:78:11:E: end tag for "section" which is
not finished
/usr/bin/openjade:cepprop.xml:70:129:X: reference to non-existent ID
"IDBackstrand2001X"
/usr/bin/openjade:cepprop.xml:70:94:X: reference to non-existent ID
"IDBackstrand2002X"
/usr/bin/openjade:cepprop.xml:70:43:X: reference to non-existent ID
"IDBackstrand2002"
/usr/bin/openjade:cepprop.xml:70:29:X: reference to non-existent ID
"IM0"
getbib error: could not create footer:

make: *** [cepprop.bib.xml] Error 1

The base xml file as well as the xsl.err, bib.xml.err and id.xml are all
created and look fine. Here's the tail-end of refdbd's debug output:

SELECT DISTINCT dbname, orig_id, author_concat, pubyear, article_title,
periodical, volume, issue, startpage, entry_id, citation_pos, xref_pos,
multi_id, sorted_pos FROM getbibtemp WHERE citation_pos=3D0 ORDER BY
xref_pos
temp table access error
could not create footer:
command processing done, finish dialog now
child finished client on fd 5
child exited with code 0

Any help?
--=20
Dylan Oliver=20
http://bork.hampshire.edu/~dao
AIM: l0sang

[Refdb-users] refdbib segfault

[Markus H. <mar...@mh...>]

Hi,

what version are you working with? I know that the last release 0.9.2a
does have a couple of problems related to citation keys. Most of these
issues have been fixed in CVS and I'd like to know whether this also
holds true for the problem you stumbled in. Could you please provide
everything necessary to reproduce the problem by private mail, i.e. a
zip of the subdirectory containing your Makefile, cepprop.xml, the err
files etc, plus the RIS file containing your references.

BTW why do you assume that refdbib segfaults? Do you get a core file?
I gather from the information that you provided that refdbd bails out
after an error which causes refdbib to call it quits.

regards,
Markus

Dylan Oliver writes:
 > Hi,
 > 
 > I started a document with refdbnd (docbook xml 4.1.2), inserted some
 > references to sources I added to my database (mysql), and have been
 > trying to compile for hours now without any luck. Here's what I'm
 > seeing: 
 > 
 > make html                                 
 > runbib -d rdbib_dao -S J.Biol.Chem. -t db31x cepprop.xml
 > /usr/bin/openjade:cepprop.xml:76:4:E: cannot find "cepprop.bib.xml";
 > tried "cepprop.bib.xml", "/usr/local/share/sgml/cepprop.bib.xml",
 > "/usr/local/lib/sgml/cepprop.bib.xml", "/usr/lib/sgml/cepprop.bib.xml"
 > /usr/bin/openjade:cepprop.xml:78:11:E: end tag for "section" which is
 > not finished
 > /usr/bin/openjade:cepprop.xml:70:129:X: reference to non-existent ID
 > "IDBackstrand2001X"
 > /usr/bin/openjade:cepprop.xml:70:94:X: reference to non-existent ID
 > "IDBackstrand2002X"
 > /usr/bin/openjade:cepprop.xml:70:43:X: reference to non-existent ID
 > "IDBackstrand2002"
 > /usr/bin/openjade:cepprop.xml:70:29:X: reference to non-existent ID
 > "IM0"
 > getbib error: could not create footer:
 > 
 > make: *** [cepprop.bib.xml] Error 1
 > 
 > The base xml file as well as the xsl.err, bib.xml.err and id.xml are all
 > created and look fine. Here's the tail-end of refdbd's debug output:
 > 
 > SELECT DISTINCT dbname, orig_id, author_concat, pubyear, article_title,
 > periodical, volume, issue, startpage, entry_id, citation_pos, xref_pos,
 > multi_id, sorted_pos FROM getbibtemp WHERE citation_pos=0 ORDER BY
 > xref_pos
 > temp table access error
 > could not create footer:
 > command processing done, finish dialog now
 > child finished client on fd 5
 > child exited with code 0
 > 
 > Any help?
 > -- 
 > Dylan Oliver 
 > http://bork.hampshire.edu/~dao
 > AIM: l0sang

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] refdbib segfault

[Markus H. <mar...@mh...>]

Hi,

I didn't run any tests yet, but I see that your RIS data still contain
the unmapped tags. They're not supposed to be imported into RefDB,
therefore you've got to strip them out (or use the proper switch of
the import filter to suppress them in the first place). Use a command
like:

grep -v "<unmapped" < cepprop.ris > cepprop.fixed.ris

to fix your existing data. Does your problem persist if you use clean
RIS data?

regards,
Markus

Dylan Oliver writes:
 > Sorry for the confusion re: segfaulting; I was running 0.9.2a when I
 > started writing that message and the segfaulting stopped when I upgraded
 > to 0.9.3-pre6 and `make clean`ed. The references were imported from
 > pubmed through RefDB's web interface. Thanks, and let me know if I can
 > be of any further assistance in debugging this.  
 > 

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] refdbib segfault

[Markus H. <mar...@mh...>]

Hi Dylan,

Dylan Oliver writes:
 > I still get the same error after removing the unmapped lines. 
 > 

Fine, then it's a bug. I'll try to run some tests with your data asap
and see what happens.

 > How about automatically stripping unmapped tags when importing data
 > through the web interface?
 > 

Sure this is possible. It never occured to me, that's why it hasn't
been done long ago. I'll add a proper default setting to
refdbcgirc.example and add some code to the RIS import routine that
will skip lines starting with "<unmapped>".

regards,
Markus

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] refdbib segfault

[Markus H. <mar...@mh...>]

Hi,

Markus Hoenicka writes:
 > Fine, then it's a bug. I'll try to run some tests with your data asap
 > and see what happens.
 > 

Done. There's been two issues:

- your XML document was broke. The bibliography entity lacked the
  trailing ";" which causes SGML/XML parsers to barf. The whole line
  needs to read:
  &bibliography;

- RefDB stumbled over the citation key case issue, something that I'm
  currently about to fix thoroughly. Your data now work properly on my
  box, but there's been a couple of changes compared to the last
  release. I'll check in the changes later tonight. You should then be
  able to pull the CVS version (if the SF folks will let you) and get
  things to work. The current version relies on xsltproc to process
  the intermediate XML documents. You better make sure you've
  installed that lest you want to manually fix the bibliography
  scripts.

 >  > How about automatically stripping unmapped tags when importing data
 >  > through the web interface?
 >  > 
 > 
 > Sure this is possible. It never occured to me, that's why it hasn't
 > been done long ago. I'll add a proper default setting to
 > refdbcgirc.example and add some code to the RIS import routine that
 > will skip lines starting with "<unmapped>".
 > 

I've changed the med2riscgirc example config file to ignore unmapped
tags. The import routines actually ignore these lines anyway, they
only result in additional warnings in the log output.

regards,
Markus

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] refdbib segfault

[Markus H. <mar...@mh...>]

Dylan Oliver writes:
 > The big secret, I discovered, is to comment out &bibliography; 
 > from short.xml before running make for the first time, to uncomment it,
 > and finally run make again. voila! Long as I searched through the docs, this
 > was NOT obvious; please spell it out in the tutorial or, better, find a
 > work-around (I tried adding '--novalid' to xsltproc's list of arguments
 > in runbib, but it didn't do the trick).
 > 
I'll use the workaround you suggested in a separate mail. I didn't
bother to fix this in the past as it is a non-fatal error that
shouldn't do any harm except that it doesn't look reassuring to start
with an error.

 > Oh, one more change for the docs: I noticed that the Y:author prefix
 > produces a reference number rather than year in the text - that is, a
 > (1) instead of a (2000). Be nice to have both options, really - maybe
 > the former could be accomplished by prefixing the key with an 'N:'. But
 > 'Y:' should really just print the year.
 > 
This is by design. The year should only be used in author/year
citation styles as far as I can tell from the journals that I read:

...as reported lately (Miller, 1999)
...as reported by Miller (1999)

...as reported lately (5)
...as reported by Miller (5)

Having both options is not a good thing as it would make switching
styles hard. Using prefixes that work only in author/year or only in
numbered citation styles is not a good thing IMHO.

regards,
Markus

-- 
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de