refdb-users — List for RefDB users

Re: [Refdb-users] slight problem with refdb 0.9.8-1 distribution

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

Hi Vince,

Quoting Vincent Carey 525-2265 <st...@ch...>:

>
> i apologize if i should be writing to some mailing list
>

Your message will reach me either way, but mailing lists have the =20
benefit of a public archive. The list information is right on =20
http://refdb.sourceforge.net. I've cc'ed my reply to the refdb-users =20
list.

>
> note the *dump.sqlite* has a .2.
>
> in line 190  of refdb-init.in in the same folder we have
>     checkFile "<pkgdatadir>/sql/refdb.dump.sqlite" "read" \
>
> and the lack of the .2. is a problem for refdb-init
>

That's me being careless... however, the 0.9.9 prereleases use a =20
different mechanism to create or update the main database. refdbd now =20
has a maintenance mode that takes care of this stuff. Therefore the =20
line above no longer exists in refdb-init. To the best of my knowledge =20
the new mechanism does not show this problem, but I'll check again to =20
make sure. I'll probably make a fixed version of the old refdb-init =20
available on the web page for those who still use 0.9.8-1.

> while i have your attention, the default behavior of the config files
> is to do logging and pid saving in dirs that are typically root controlled=
.
> it would be nice to make these easily configurable, living under prefix va=
lue
> by default.
>

The log destinations, log files, and pid files are configurable in the =20
init files (<prefix>/etc/refdb/refdbXrc) and via the command line (see =20
the -e, -L, and -P switches). I agree that the default locations =20
should include the prefix, although I think that on most systems both =20
/var/log and /usr/local/var/log are root controlled. I'll add the =20
prefix asap.

regards,
Markus

--=20
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

[Refdb-users] Strange behaviour of addref called from shell script (FIXED)

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

Hi,

I've been working on this problem off-line with Jeremy (thanks for
the ssh access!). Here's a brief summary:

- there were a few sprintf commands that did not work around an OS X
  sprintf bug which I reported a while ago. All these commands were
  fixed
- the real problem for the apparent insert failure was a
  missing prototype for a function which returns a long long
  value. The compiler assumes an 'int' as a return value in this
  case. This is not a problem on Intel/AMD boxes as the significant
  bits are in the same position on the stack, but a PPC processor
  looks at the 'wrong end' of the value and returns zero instead of an
  ID value. Adding the prototype to refdbd.h.in fixes this problem.

I've checked in the fixed header file. An updated prerelease
containing this fix will be available shortly.

regards,
Markus

Jeremy Malcolm writes:
 > Why might refdbd be showing this error (multiplied by 837):
 > 
 > BEGIN
 > LOCK TABLES t_refdb WRITE, t_author WRITE, t_keyword WRITE, t_periodical 
 > WRITE, t_user WRITE, t_xauthor WRITE, t_xkeyword WRITE, t_xuser WRITE, 
 > t_note WRITE, t_xnote WRITE, t_link WRITE, t_xlink WRITE, 
 > refdb.t_journal_words WRITE
 > SELECT refdb_id, refdb_citekey FROM t_refdb WHERE 
 > refdb_citekey='DIJK2000MODELS-OF-DEMOC'
 > INSERT INTO t_refdb (refdb_type,refdb_citekey) VALUES 
 > ('DUMMY','DUMMYJeremy-Malcolms-Computer.local11086')
 > 1062: Duplicate entry 'DUMMYJeremy-Malcolms-Computer.local11086' for key 2
 > inserting reference data failed
 > ROLLBACK
 > UNLOCK TABLES
 > failed processing dataset
 > 

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

Re: [Refdb-users] Strange behaviour of addref called from shell script

[Jeremy M. <Je...@Ma...>]

Markus Hoenicka wrote:
> There is indeed nothing unusual here. In fact, I use a (far from
> perfect) test harness here which is a shell script like yours. It
> creates databases, inserts, updates, picks, unpicks, deletes
> references and notes. I have never seen any difference between running
> a command from the command line or from within a script.

Suddenly, it just started doing the same thing from the command line. 
Then I exited from a few applications and tried again and it worked.  So 
I guess the problem wasn't really caused by running refdbc from a 
script, it was just that the extra overhead of another shell instance 
created a lack of system resources.

In any case, when it fails, addref produces this output:

233:1
414:1
233:2
414:2
233:3
414:3
233:4
414:4
233:5
414:5 ...

Whereas when it succeeds it produces output like this (showing my 
citation keys):

406:1:RFC3492
408:1
406:2:GAC2006GAC-PRINCIPLES-
408:2
406:3:SMITH2001TAKING-DELIBERA
408:3
406:4:NYGH1995CONFLICT-OF-LAW
408:4
406:5:MUELLER2006A-FUNNY-THING-H
408:5
406:6:MUELLER2005WHAT-TO-DO-ABOU ...

> The only way to bump into this problem is to cause refdbd to stop
> inserting data before it updates the citation key of a newly created
> dataset. This would create an incomplete dataset with that particular
> dummy citation key. This problem also requires lack of transaction
> support, as otherwise the rollback (or the missing commit in case of a
> crash) would cause MySQL to remove the dataset again. I don't see how
> running refdbc from a script could cause this problem, but it seems to
> be a fact. Could you please do the following:
> 
> - check the MySQL log for additional hints
> - create a SQL dump of your thesis database after your script ran
>   unsuccessfully. Please include the refdbd log of that particular run
>   as I'd like to compare the process IDs involved.

Sending MySQL log, database dump and refdbd log privately.

-- 
Jeremy Malcolm LLB (Hons) B Com
Internet and Open Source lawyer, IT consultant, actor
host -t NAPTR 1.0.8.0.3.1.2.9.8.1.6.e164.org|awk -F! '{print $3}'

Re: [Refdb-users] Strange behaviour of addref called from shell script

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

Jeremy Malcolm writes:
 > #!/bin/bash
 > refdba -c /bin/cat -C deletedb thesis
 > cat Thesis.bib | bib2xml -c corporations.txt | xml2ris | sed -e 
 > '/^AU.*/s/\.//g; s/STD$/ELEC/' > Thesis.ris
 > # Capitalise references, to work around a bug in openjade
 > runsed lyxtox/sedscr_ris Thesis.ris
 > refdba -c /bin/cat -C createdb thesis -E latin1
 > refdbc -c /bin/cat -d thesis -C addref Thesis.ris
 > 

There is indeed nothing unusual here. In fact, I use a (far from
perfect) test harness here which is a shell script like yours. It
creates databases, inserts, updates, picks, unpicks, deletes
references and notes. I have never seen any difference between running
a command from the command line or from within a script.

 > Using updateref doesn't seem to produce the same problem, however I need 
 > to use addref because I will usually have a mixture of new, old and 
 > (possibly) updated references to import.
 > 

This is correct. updateref bypasses the "INSERT" command as it should
touch only existing datasets. This way you don't bump into the
duplicate problem.

 > 0.97 I think.  Running on Mac OS X 10.3.9.
 > 

Oh dear. There have been way too many changes between 0.9.7 and the
latest prerelease.

The only way to bump into this problem is to cause refdbd to stop
inserting data before it updates the citation key of a newly created
dataset. This would create an incomplete dataset with that particular
dummy citation key. This problem also requires lack of transaction
support, as otherwise the rollback (or the missing commit in case of a
crash) would cause MySQL to remove the dataset again. I don't see how
running refdbc from a script could cause this problem, but it seems to
be a fact. Could you please do the following:

- check the MySQL log for additional hints
- create a SQL dump of your thesis database after your script ran
  unsuccessfully. Please include the refdbd log of that particular run
  as I'd like to compare the process IDs involved.

regards,
Markus

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

Re: [Refdb-users] Strange behaviour of addref called from shell script

[Jeremy M. <Je...@Ma...>]

Markus Hoenicka wrote:
> Two questions:
> 
> - what does the shell script do? Could you please post it here or send  
> it to me off-list?

It is much longer than this, but here is the start of it - the last line 
is the one that produces the error when scripted.

#!/bin/bash
refdba -c /bin/cat -C deletedb thesis
cat Thesis.bib | bib2xml -c corporations.txt | xml2ris | sed -e 
'/^AU.*/s/\.//g; s/STD$/ELEC/' > Thesis.ris
# Capitalise references, to work around a bug in openjade
runsed lyxtox/sedscr_ris Thesis.ris
refdba -c /bin/cat -C createdb thesis -E latin1
refdbc -c /bin/cat -d thesis -C addref Thesis.ris

Using updateref doesn't seem to produce the same problem, however I need 
to use addref because I will usually have a mixture of new, old and 
(possibly) updated references to import.

> - which version did you upgrade from? Knowing that will help me to  
> locate the problem.

0.97 I think.  Running on Mac OS X 10.3.9.

Thanks!

-- 
Jeremy Malcolm LLB (Hons) B Com
Internet and Open Source lawyer, IT consultant, actor
host -t NAPTR 1.0.8.0.3.1.2.9.8.1.6.e164.org|awk -F! '{print $3}'

Re: [Refdb-users] Strange behaviour of addref called from shell script

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

Quoting Jeremy Malcolm <Je...@Ma...>:

> Why might refdbd be showing this error (multiplied by 837):
>
[...]

> when I try to add my references using a shell script, but NOT when I run
> the exact same command from the bash prompt?  My command is refdbc -c
> /bin/cat -d thesis -C addref Thesis.ris, and this problem has only
> arisen since the most recent upgrade to 0.9.9-pre2.
>

Two questions:

- what does the shell script do? Could you please post it here or send  
it to me off-list?
- which version did you upgrade from? Knowing that will help me to  
locate the problem.

regards,
Markus

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

[Refdb-users] Strange behaviour of addref called from shell script

[Jeremy M. <Je...@Ma...>]

Why might refdbd be showing this error (multiplied by 837):

BEGIN
LOCK TABLES t_refdb WRITE, t_author WRITE, t_keyword WRITE, t_periodical 
WRITE, t_user WRITE, t_xauthor WRITE, t_xkeyword WRITE, t_xuser WRITE, 
t_note WRITE, t_xnote WRITE, t_link WRITE, t_xlink WRITE, 
refdb.t_journal_words WRITE
SELECT refdb_id, refdb_citekey FROM t_refdb WHERE 
refdb_citekey='DIJK2000MODELS-OF-DEMOC'
INSERT INTO t_refdb (refdb_type,refdb_citekey) VALUES 
('DUMMY','DUMMYJeremy-Malcolms-Computer.local11086')
1062: Duplicate entry 'DUMMYJeremy-Malcolms-Computer.local11086' for key 2
inserting reference data failed
ROLLBACK
UNLOCK TABLES
failed processing dataset

when I try to add my references using a shell script, but NOT when I run 
the exact same command from the bash prompt?  My command is refdbc -c 
/bin/cat -d thesis -C addref Thesis.ris, and this problem has only 
arisen since the most recent upgrade to 0.9.9-pre2.

-- 
Jeremy Malcolm LLB (Hons) B Com
Internet and Open Source lawyer, IT consultant, actor
host -t NAPTR 1.0.8.0.3.1.2.9.8.1.6.e164.org|awk -F! '{print $3}'

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[Jeremy M. <Je...@Ma...>]

Markus Hoenicka wrote:
> Jeremy Malcolm writes:
>  > Anyway, I'm sure it is my misunderstanding, but the reason why I do it 
>  > that way is because I need RefDB to print (ed) or (eds) after their 
>  > names if they are editors, and it won't do that unless I put them into 
>  > A2 rather than A1.
> 
> Please have a look at 0.9.9-pre2 if you find some time. I'd like to
> know whether the most recent changes to the bibliography sorting stuff
> also fix your problem.

Yes!  Perfect now.  Thank you!

-- 
Jeremy Malcolm LLB (Hons) B Com
Internet and Open Source lawyer, IT consultant, actor
host -t NAPTR 1.0.8.0.3.1.2.9.8.1.6.e164.org|awk -F! '{print $3}'

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

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

Jeremy Malcolm writes:
 > Anyway, I'm sure it is my misunderstanding, but the reason why I do it 
 > that way is because I need RefDB to print (ed) or (eds) after their 
 > names if they are editors, and it won't do that unless I put them into 
 > A2 rather than A1.
 > 

Please have a look at 0.9.9-pre2 if you find some time. I'd like to
know whether the most recent changes to the bibliography sorting stuff
also fix your problem.

regards,
Markus

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

[Refdb-users] ANN: refdb-0.9.9-pre2 available

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

Hi,

this prerelease addresses some of the problems related to bibliography
sorting. Bibliographic entries are now sorted in a case-insensitive
fashion. The title is used for sorting if the entry does not have an
author (except when the style uses something like "Anonymous" instead
of an author name). Finally, "ALL" (which actually means "first
available") is supported in bibliography styles as an attribute value
for author lists in addition to "PRIMARY" through "TERTIARY". This
allows RefDB to use secondary authors e.g. in books if the datasets do
not contain primary authors.

The prerelease is available here:

http://refdb.sourceforge.net/pre/refdb-latest.tar.gz

As mentioned in the previous announcement, the current prereleases
require updated Perl modules if you rely on their functionality:

http://refdb.sourceforge.net/pre/RefDB-perlmod-1.2.tar.gz

http://refdb.sourceforge.net/pre/RefDB-Client-1.17.tar.gz

http://refdb.sourceforge.net/pre/RefDB-SRU-0.6.tar.gz

regards,
Markus


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

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[Bruce D'A. <bda...@gm...>]

On 4/23/07, Jus...@pi... <Jus...@pi...> wrote:
> "Bruce D'Arcus" <bda...@gm...> wrote on Mon, 23 Apr 2007
> 10:34:11 -0400:
>
> > But sorting conventions vary by locale,
>
> What I meant and neglected to say is that sorting conventions are
> usually imposed by the required bibliography style, and that the
> sorting locale should thus be set by the bibliography style
> definition.

True, but sorting still depends on the name in question. If I have a
style that says sort on "last name" in English, "Mao Zedong" still
sorts on the first name, which is the family name.

> > I've come around to believing that it's easier and more
> > straightforward for people records to have an explicit sort-string
> > property, as vCard does.
>
> I would love this to be true, but is it? Does the entire world agree
> on how to sort names beginning with Van, von, ... (D'...)?

But here's the complication as I understand it. The sorting convention
of articulars is not just a question of the display locale, but of the
origin of the name itself. The last I looked into this, some European
languages would sort on the articulars, and others not.  And I presume
that they might be mixed in a bibliography list.

Note: I'm not 100% sure of this and would like to clarify, but
certainly doing simple culturally-specific name parts like first and
last is not at all international-friendly. And even family, given ,
etc. the introduces the issues of how you sort different kinds of
names.

Bruce

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

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

Quoting Jeremy Malcolm <Je...@Ma...>:

> Anyway, I'm sure it is my misunderstanding, but the reason why I do it
> that way is because I need RefDB to print (ed) or (eds) after their
> names if they are editors, and it won't do that unless I put them into
> A2 rather than A1.
>

I see. This is a part of the RIS hell that needs to be fixed with an  
extended data model.

I've started working on these issues. Sorting "van Beethoven" before  
"ZZ Top" seems to work now, but I haven't got round to address the  
missing A1 in BOOKs issue yet. This will all show up in the next  
prerelease.

regards,
Markus

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

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[Jeremy M. <Je...@Ma...>]

 > Quoting Jeremy Malcolm <Jeremy <at> Malcolm.id.au>:
 >
 > > Another problem is that if no primary author is specified but a
 > > secondary author is specified (generally because you are citing a
 > > book that has editor/s rather than author/s), it lists all of
 > > those references at the start of the bibliography, before the letter
 > > A.  The preferred behaviour would obviously be to interleave them
 > > with the other references.
 >
 > This is probably due to my (or your?) misunderstanding of how a book
 > should be encoded in RIS. I've assumed that the names printed on the
 > cover go into A1.

Firstly, sorry for the delay in following up.  I had disabled mail 
delivery from this list for a while and thought I had re-enabled it, but 
Sourceforge is a bit flaky that way.

Anyway, I'm sure it is my misunderstanding, but the reason why I do it 
that way is because I need RefDB to print (ed) or (eds) after their 
names if they are editors, and it won't do that unless I put them into 
A2 rather than A1.

-- 
Jeremy Malcolm LLB (Hons) B Com
Internet and Open Source lawyer, IT consultant, actor
host -t NAPTR 1.0.8.0.3.1.2.9.8.1.6.e164.org|awk -F! '{print $3}'

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[<Jus...@Pi...>]

"Bruce D'Arcus" <bda...@gm...> wrote on Mon, 23 Apr 2007
10:34:11 -0400:

> But sorting conventions vary by locale,

What I meant and neglected to say is that sorting conventions are
usually imposed by the required bibliography style, and that the
sorting locale should thus be set by the bibliography style
definition.

> I've come around to believing that it's easier and more
> straightforward for people records to have an explicit sort-string
> property, as vCard does.

I would love this to be true, but is it? Does the entire world agree
on how to sort names beginning with Van, von, ... (D'...)?

Justus

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[Bruce D'A. <bda...@gm...>]

On 4/17/07, Jus...@pi... <Jus...@pi...> wrote:
> Markus Hoenicka <mar...@mh...> wrote on Tue, 17 Apr
> 2007 15:16:27 +0200:
>
> > is it reasonable to expect e.g. from an XSLT-based bibliography
> > formatting system to correctly identify the first word or words
> > relevant for sorting names or titles?
>
> For names, I think the proper way would be to represent their
> different parts separately (as BibTeX does it implictly through
> obscure syntax conventions); they could then be accessed individually
> by style sheets.

But sorting conventions vary by locale, so you can't rely on this.

In Asia (and indeed, in Western Europe when dealing with Asian names),
for example, you sort the same way you display. E.g. "Mao Zedong"
sorts like "Mao Zedong."

I've come around to believing that it's easier and more
straightforward for people records to have an explicit sort-string
property, as vCard does.

Bruce

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

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

Quoting Markus Hoenicka <mar...@mh...>:

> cover go into A1. Compare with book chapter: The author of the chapter
> goes into A1, the book title into T1, and the book editors into A2.

please read as:

> cover go into A1. Compare with book chapter: The author of the chapter
> goes into A1, the book title into *T2*, and the book editors into A2.

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

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

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

Quoting Jeremy Malcolm <Je...@Ma...>:

> Another problem is that if no primary author is specified but a
> secondary author is specified (generally because you are citing a book
> that has editor/s rather than author/s), it lists all of those
> references at the start of the bibliography, before the letter A.  The
> preferred behaviour would obviously be to interleave them with the other
> references.
>

This is probably due to my (or your?) misunderstanding of how a book =20
should be encoded in RIS. I've assumed that the names printed on the =20
cover go into A1. Compare with book chapter: The author of the chapter =20
goes into A1, the book title into T1, and the book editors into A2. =20
Now if the book title goes into T1 in a whole book reference, I =20
assumed it to be straightforward to have the names in A1, regardless =20
of whether they act as authors or editors. You apparently prefer to =20
put authors into A1 and editors into A2. This is the usual mess caused =20
by RIS mixing up the orthogonal concepts of bibliographic levels and a =20
person's responsibility.

I'm sure that a more sane data format will allow to avoid this in the =20
future, but for the time being I'll try to have RefDB check for =20
missing primary authors and use secondary ones whenever this makes =20
sense.

regards,
Markus



--=20
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[Jeremy M. <Je...@Ma...>]

Jeremy Malcolm wrote:
> It there any way for RefDB not to sort bibliography entries in this way?
> 
> Zippay, A ...
> Zittrain, Jonathan ...
> Zumbansen, P ...
> de Montesquieu, Charles ...
> de la Chapelle, Bertrand ...
> van Dijk, Jan ...
> von Hippel, Eric ...

Another problem is that if no primary author is specified but a 
secondary author is specified (generally because you are citing a book 
that has editor/s rather than author/s), it lists all of those 
references at the start of the bibliography, before the letter A.  The 
preferred behaviour would obviously be to interleave them with the other 
references.

-- 
Jeremy Malcolm LLB (Hons) B Com
Internet and Open Source lawyer, IT consultant, actor
host -t NAPTR 1.0.8.0.3.1.2.9.8.1.6.e164.org|awk -F! '{print $3}'

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[<Jus...@Pi...>]

Markus Hoenicka <mar...@mh...> wrote on Tue, 17 Apr
2007 15:16:27 +0200:

> is it reasonable to expect e.g. from an XSLT-based bibliography  
> formatting system to correctly identify the first word or words  
> relevant for sorting names or titles?

For names, I think the proper way would be to represent their
different parts separately (as BibTeX does it implictly through
obscure syntax conventions); they could then be accessed individually
by style sheets.

For titles, one would have to add markup to identify the sort-key
portion, as you said. (But who would want to sort by titles anyway :-/.)

Justus

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

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

Quoting Jus...@Pi...:

> Not sure about this... As you said, the following issues arise in
> sorting:
>
> 1. multi-part keys and their respective influence on sorting order
>    ("van Dijk", "The Title of This Article"),
>
> 2. character encoding,
>
> 3. localization.
>
> The first and third issues depend on the context (language,
> bibliography style, ...) and thus cannot be resolved by sort-key
> normalization.
>

I see. An umlaut in an English document needs a normalization to sort =20
properly, whereas the same umlaut in a German document doesn't. OTOH =20
is it reasonable to expect e.g. from an XSLT-based bibliography =20
formatting system to correctly identify the first word or words =20
relevant for sorting names or titles? In any language? I somehow feel =20
that this should be done by a human.

> The second and third issues are addressed simultaneously by the POSIX
> localization mechanism.
>
> Thus, I think bibliography sorting should, if possible, be done by
> POSIX-compliant sorting (LC_COLLATE, sort(1)) of keys normalized by
> the bibliography style.

XSLT apparently has all provisions to do a POSIX-compliant sorting, =20
whereas DSSSL almost certainly doesn't (so I'm screwed again, for the =20
time being).

regards,
Markus


--=20
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[<Jus...@Pi...>]

Markus Hoenicka <mar...@mh...> wrote on Tue, 17 Apr
2007 10:37:49 +0200:

> we'd have to add a field that records a normalized name string (case
> insensitive, with umlauts and accents replaced in an intelligent
> fashion) to provide a real solution to the sorting problem.

Not sure about this... As you said, the following issues arise in
sorting:

1. multi-part keys and their respective influence on sorting order
   ("van Dijk", "The Title of This Article"),

2. character encoding,

3. localization.

The first and third issues depend on the context (language,
bibliography style, ...) and thus cannot be resolved by sort-key
normalization.

The second and third issues are addressed simultaneously by the POSIX
localization mechanism.

Thus, I think bibliography sorting should, if possible, be done by
POSIX-compliant sorting (LC_COLLATE, sort(1)) of keys normalized by
the bibliography style.

Justus

Re: [Refdb-users] RefDB and Jurabib

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

Hi Doug,

Quoting Doug du Boulay <dou...@gm...>:

>
> As I recall, that was a route to RTF format no? Maybe now with
> the MS word and OpenOffice.org XML formats there may be other
> pure XSLT based avenues?
>

I've been investigating one of them =20
(http://open.comsultia.com/docbook2odf/) which looks quite promising. =20
This is essentially a Perl control script that creates the files which =20
are then zipped up to the .odf file, plus a set of stylesheets that do =20
the actual XML-to-XML transformations. Unfortunately the project =20
currently does not provide any templates for the <citation> or =20
<bibliography> elements, so we'd have to write this from scratch. I'd =20
be more than happy if someone investigated this path. I'm afraid I =20
won't get around in the near future.

> It was written at a time when I had toooo much free time on my hands.
> But I would be prepared to spend some more time  on it, and happy to try a=
nd
> explain the inner workings if you were of a mind to adapt or recode to som=
e
> other style/purpose.
>

Much appreciated.

> Basically the way it works was to suck all the references out of  =20
> RefDB in RISX
> format, then XSLT transform them to my arbitrary DocBook raw format
> (about 5% of the code),
> then format DocBook raw according to your citestylex specs (about 95%
> of the code). It also works fine with a separate DocBook raw
> bibliography file.  The XSLT to extract and organise the RISX could be
> reused to make it a distinct two step process,
> saving DocBook raw to a separate file and then applying formatting  =20
> afterwards.
> (though I recall that my DocBook raw intermediate format may have had some
> shortcomings).
>
> This also means that opportunities for code sharing with a TEI output opti=
on
> may be quite limited ( I have never even looked at TEI to see if there are
> similar raw bibliography structuring conventions).
>

Things are a little better these days. First of all, the SRU interface =20
could be extended to all other XML formats (currently only MODS and =20
risx are supported), so you could retrieve them e.g. as raw DocBook as =20
well. RefDB has always had the option to export to DocBook directly.

Next, the RefDB XML output can use a namespace these days. You could =20
essentially use a namespaced risx (or MODS, or DocBook) raw =20
bibliography and transform that independent of the surrounding =20
document format (DocBook, TEI, or whatever).

As for the TEI P4 bibliography format, the structure is more similar =20
to risx than to DocBook. TEI P5 will be a different story though, it =20
is said to be closer to MODS. But it is probably a better approach to =20
use just one namespaced format instead of transforming it to an =20
intermediate DocBook|TEI bibliography.

regards,
Markus

--=20
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

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

Hi Jeremy,

Quoting Jeremy Malcolm <Je...@Ma...>:

> It there any way for RefDB not to sort bibliography entries in this way?
>
> Zippay, A ...
> Zittrain, Jonathan ...
> Zumbansen, P ...
> de Montesquieu, Charles ...
> de la Chapelle, Bertrand ...
> van Dijk, Jan ...
> von Hippel, Eric ...
>

There is no such way out of the box, but I see that we need one. Would =20
you be kind enough to create a feature request on SourceForge?

I see two solutions to this problem. The more limited approach is to =20
make the ORDER BY case-insensitive (SQLite and newer versions of MySQL =20
allow this via collations; PostgreSQL apparently requires uppercasing =20
or something). I guess this could be part of the next prerelease, =20
unless I miss some basic problem with this approach.

However, this is may be insufficient to solve sorting problems with =20
special characters like umlauts or accented characters, as these are =20
often sorted outside a-z. Therefore, some bibliographic data formats =20
like MODS carry extra fields that describe where a name or a title =20
should appear in a sorted list. Just like titles are often sorted by =20
the first non-article word instead of just lexically (ain't no fun to =20
scan through half a million titles starting with "The"), the compound =20
name "van Dijk" in your example can appear either close to =20
"Dijkstra,A." or close to "Vandijk,A.". I'm sure this distinction =20
depends on the bibliography style as well. In any case, we'd have to =20
add a field that records a normalized name string (case insensitive, =20
with umlauts and accents replaced in an intelligent fashion) to =20
provide a real solution to the sorting problem.

regards,
Markus


--=20
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

Re: [Refdb-users] troubles with phpweb setup

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

Quoting Cecile Hebert <cec...@tu...>:

>
> It works, I just have to call
> http://tem.atp.tuwien.ac.at/refdb/
> *WITH* the trailing slash. Does not work without.
>

This is correct, but the phpweb/INSTALL file already mentions this.

> I am not sure this is the most "academic" way, but at least I get the
> pages... (somehow I feel the symbolic linkt should not be necessary...)
>

It certainly isn't necessary in my case. I wonder if this is an Apache =20
1.3 vs. 2.0 issue.

> [Mon Apr 16 23:06:32 2007] [error] [client 195.230.58.82] PHP Notice:
> Undefined variable: passwd in /usr/local/share/refdb/www/login.php on
> line 6, referer: http://tem.atp.tuwien.ac.at/refdb/index.html
>

PHP calls this a "Notice". I'm not sure whether it really hurts.

> So I would say php seems to run ?
>

Yes it does. I was just asking because I had accidentally upgraded =20
php5 without upgrading php-session. In my case php5 refused to run =20
because it couldn't load php-session.

regards,
Markus

--=20
Markus Hoenicka
mar...@ca...
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

[Refdb-users] citestyle's BIBLIOSEQUENCE sorts upper and lower alpha separately

[Jeremy M. <Je...@Ma...>]

It there any way for RefDB not to sort bibliography entries in this way?

Zippay, A ...
Zittrain, Jonathan ...
Zumbansen, P ...
de Montesquieu, Charles ...
de la Chapelle, Bertrand ...
van Dijk, Jan ...
von Hippel, Eric ...

TIA

-- 
Jeremy Malcolm LLB (Hons) B Com
Internet and Open Source lawyer, IT consultant, actor
host -t NAPTR 1.0.8.0.3.1.2.9.8.1.6.e164.org|awk -F! '{print $3}'