refdb-users — List for RefDB users

[Refdb-users] unscramble string

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

Hi,

this bug was reported previously, and it's fixed in 0.9.4-pre1 which I
recommend anyway as it fixes a boatload of other bugs:

http://refdb.sourceforge.net/pre/refdb-0.9.4-pre1.tar.gz

regards,
Markus

rob caSSon writes:
 > i was getting the "unscramble string" segfault, and it only seems to
 > happen when i'm logging to STDERR....when i log to a file, all seems to
 > work well...
 > 
 > just thought others might find this tidbit useful....
 > 
 > btw: on debian gnu/linux stable...
 > 

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

[Refdb-users] web client idea

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

Bruce D'Arcus writes:
 > 
 > A small idea for the current web client:
 > 
 > When adding a new RIS reference, it would be nice if by default an 
 > empty template was inserted.  It's often impossible for me to remember 
 > the tags, and a problem if I'm away from my machine with its emacs + 
 > ris mode (as I was a little while ago) and want to enter a record.
 > 

Ceterum censeo ... this should go into a Perl CGI program. I loathe to
fiddle with the current CGI implementation (except bug fixes) as I
know I'll have to dump it anyway.

regards,
Markus

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

[Refdb-users] unscramble string

[rob c. <ro...@st...>]

i was getting the "unscramble string" segfault, and it only seems to
happen when i'm logging to STDERR....when i log to a file, all seems to
work well...

just thought others might find this tidbit useful....

btw: on debian gnu/linux stable...

[Refdb-users] hmm...koha?

[Bruce D'A. <bd...@fa...>]

Question:

If I see the following in html source, does this mean the web-client is 
using perl?

loadmodules.pl

If, as I assume, yes, then it seems that the open source MySQL-based 
Koha library application is using perl for its web client.  Is there 
any chance this could be useful for a new refdb web client that uses 
the perl module?  It is a library application, but there is, for 
example, an interface for entering new records.  You can either search 
and import, or also manually enter new MARC records (thought the UI for 
this is a little confusing).

More (with demos) at:

http://www.koha.org/

Bruce

[Refdb-users] web client idea

[Bruce D'A. <bd...@fa...>]

A small idea for the current web client:

When adding a new RIS reference, it would be nice if by default an 
empty template was inserted.  It's often impossible for me to remember 
the tags, and a problem if I'm away from my machine with its emacs + 
ris mode (as I was a little while ago) and want to enter a record.

Bruce

[Refdb-users] running refdbd as a Windows service?

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

Hi,

I've been playing with RefDB on a Windows XP box lately. Essentially
everything works out of the box, except when I try to run refdbd as a
Windows service. When I start it manually with the refdbctl script,
everything works as expected.

The service is started properly, but the first query from a client
always times out. The next one usually succeeds, and things are fine
if I keep chugging on. After a break of say five minutes, the first
query will again time out. Did anyone else notice this odd behaviour?
Is it specific to WinXP or is that just bad luck for me?

Thanks
Markus

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

Re: [Refdb-users] adding reference failure

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

Bruce D'Arcus writes:
 > I think you should, because you'll need to at some point anyway. =20=

 > Consider that those styles that make a distinction between first and=
=20
 > subsequent citations almost always use abbreviated titles for the=20=

 > latter (can you imagine not doing so with the title example I posted=
=20
 > the other day?).
 >=20

As a life scientist I can't imagine citing titles anyway. On this side
of the divide we use numeric or author/year.

 > Does the current version of RefMan really not support an abbreviated=
=20
 > title field?  If yes, that surprises me.
 >=20

Trust me. I think the program was developed with life scientists in
mind...

 > Before you contemplate if and how to add this, I'd take a look at ho=
w=20
 > MODS handles titles to make sure you cover it all.  IIRC, they also=20=

 > have an attribute for "uniform" titles, which AFAIK is simply the fu=
ll=20
 > title.  So you can have:
 >=20
 > <titleInfo type=3D"uniform">
 > =09<title>A Full Title: With Subtitle</title>
 > <titleInfo>
 >=20
 > or...
 >=20
 > <titleInfo type=3D"abbreviated">=A0
 > =09<title>Annu. rep. notif. dis.</title>
 > </titleInfo>
 >=20

I'll have a look at MODS again before going ahead.

regards,
Markus

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

Re: [Refdb-users] adding reference failure

[Bruce D'A. <bd...@fa...>]

On Tuesday, September 23, 2003, at 05:44  PM, Markus Hoenicka wrote:

> Bruce D'Arcus writes:
>> But isn't there something like an abbreviated title field?  If yes, I
>> was wondering if it made sense to have the title minus the subtitle =
in
>> that field.  Maybe better than nothing?
>
> There ain't. Unfortunately. As I'm currently fiddling with the
> database schema anyway, I *could* add an abbrev title field and a
> subtitle field, but they would be useless for RIS data (and hence, for
> the web interface as it is). risx would need to be amended to support
> this.

I think you should, because you'll need to at some point anyway. =20
Consider that those styles that make a distinction between first and=20
subsequent citations almost always use abbreviated titles for the=20
latter (can you imagine not doing so with the title example I posted=20
the other day?).

Does the current version of RefMan really not support an abbreviated=20
title field?  If yes, that surprises me.

Before you contemplate if and how to add this, I'd take a look at how=20
MODS handles titles to make sure you cover it all.  IIRC, they also=20
have an attribute for "uniform" titles, which AFAIK is simply the full=20=

title.  So you can have:

<titleInfo type=3D"uniform">
	<title>A Full Title: With Subtitle</title>
<titleInfo>

or...

<titleInfo type=3D"abbreviated">=A0
	<title>Annu. rep. notif. dis.</title>
</titleInfo>

Bruce=

Re: [Refdb-users] adding reference failure

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

Bruce D'Arcus writes:
 > But isn't there something like an abbreviated title field?  If yes, I 
 > was wondering if it made sense to have the title minus the subtitle in 
 > that field.  Maybe better than nothing?
 > 

There ain't. Unfortunately. As I'm currently fiddling with the
database schema anyway, I *could* add an abbrev title field and a
subtitle field, but they would be useless for RIS data (and hence, for
the web interface as it is). risx would need to be amended to support
this.

regards,
Markus

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

Re: [Refdb-users] adding reference failure

[Bruce D'A. <bd...@fa...>]

On Monday, September 22, 2003, at 05:33  PM, Markus Hoenicka wrote:

>> Which brings up a question:
>> how should I deal with titles and subtitles, since they aren't
>> supported explicitly, but are valuable to keep separate.
>
> Currently there is no blessed way to deal with this as RIS doesn't
> know subtitles.

But isn't there something like an abbreviated title field?  If yes, I 
was wondering if it made sense to have the title minus the subtitle in 
that field.  Maybe better than nothing?

Bruce

[Refdb-users] adding reference failure

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

Bruce D'Arcus writes:
 > 
 > Trying to add the following reference via the web client gets me a 
 > failure.  Why?
 > 
 > I hope the answer isn't a too-long title.

I can't tell. Did you check the logs for a hint? What was the message
you got back?

The string length shouldn't be a problem. The database tables do not
limit the length of titles so even a couple of MB should be ok.

 > Which brings up a question: 
 > how should I deal with titles and subtitles, since they aren't 
 > supported explicitly, but are valuable to keep separate.

Currently there is no blessed way to deal with this as RIS doesn't
know subtitles.

regards,
Markus

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

[Refdb-users] hmm...reference was added

[Bruce D'A. <bd...@fa...>]

Well, despite getting the failure message on adding the reference, it 
is in fact in the DB.  Go figure...

Bruce

[Refdb-users] adding reference failure

[Bruce D'A. <bd...@fa...>]

Trying to add the following reference via the web client gets me a 
failure.  Why?

I hope the answer isn't a too-long title.  Which brings up a question: 
how should I deal with titles and subtitles, since they aren't 
supported explicitly, but are valuable to keep separate.

Bruce

TY  - BOOK
ID  - Chang2002a
AU  - Chang, Nancy
PY  - 2002
BT  - Silencing Political Dissent: How Post-September 11 Anti-Terrorism 
Measures Threaten Our Liberties
CY  - New York
PB  - Seven Stories Press
KW  - COINTELPRO
KW  - protest
KW  - post-9/11
KW  - PATRIOT Act
ER  -

[Refdb-users] web query

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

Hi Bruce,

Bruce D'Arcus writes:
 > 
 > I kind expect to be able to type a partial author and have the web 
 > client return results.

Rightfully so...

 >  I want to type just "Doe" rather than "Doe, 
 > John Q."  In the absence of this, how do I get this behavior?  I can't 
 > seem to figure it out from the docs.
 > 

I assume you're using SQLite as the db engine? This is unfortunately
one of the weak spots where the implementation of the database engine
shines through. Just to make it clear, this has nothing to do with the
web interface. The command line client shows exactly the same
behaviour.

Both MySQL and PostgreSQL support RLIKE (regexp like)
comparisons. They work as expected. If you search for an author with
the string "Doe", you'll get "Doe, John Q." as well as "DiDoe,M." or
any other author whose name contains the character
sequence"Doe". SQLite does not support RLIKE, so refdbd has to use
LIKE instead. This one uses the much simpler SQL regexps and does not
perform partial matches by default. In order to find all Does in your
database, you'll have to search for "Doe%" which means "Doe" followed
by zero or more arbitrary characters. This is a minor inconvenience
when using SQLite.

BTW these nasty details are covered in the sections "The query
language" and "Regular expressions" in chapter 13 of the manual.

regards,

Markus

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

[Refdb-users] web query

[Bruce D'A. <bd...@fa...>]

I kind expect to be able to type a partial author and have the web 
client return results.  I want to type just "Doe" rather than "Doe, 
John Q."  In the absence of this, how do I get this behavior?  I can't 
seem to figure it out from the docs.

Bruce

Re: [Refdb-users] corporate names

[Bruce D'A. <bd...@fa...>]

On Monday, September 15, 2003, at 03:24  PM, Markus Hoenicka wrote:

> Hi Bruce,
>
> I assume you're talking about risx export (vs. support)?. The content
> model in the DTD looks like:
>
> <!ELEMENT author (name|(lastname?,firstname?,middlename*,suffix?))>

Yeah, I caught that later.

> so if you write risx datasets you can use the name element for
> corporate names and such. However, names are currently handled as
> strings internally, so the export routines do not have the faintest
> idea that a name once was a corporate name. This is a shortcoming of
> the current data model which can be fixed. I assume that MODS has a
> similar if not better content model for author names so the effort
> wouldn't be wasted.

Well, MODS does something like this:

<name type="corporate">
	<namePart>Blah Blah Blah</namePart>
</name>

In my experiments with a new -- more input-friendly -- schema, I was 
thinking of:

<name type="corporate">
	<fullname>Blah Blah Blah</fullname>
</name>


> I've been working on improved support for RIS as understood by
> RefMan10. This will require a change of the database schema anyway,
> and this may be an appropriate time to fix the name handling as well.
>
> regards,
> Markus
>
> Bruce D'Arcus writes:
>>
>> I just realized something.  The risx support has no concept of
>> corporate names, which seems a problem to me.  Example:
>>
>> <author><lastname>U.S. Senate Select Committee to Study Governmental
>> Operations with Respect to Intelligence Activities</lastname></author>
>>
>> I realize ris/refman likewise has no such concept, but I wonder if we
>> can say if there is only one 'name' then the following might be 
>> better:
>>
>> <author><fullname>U.S. Senate Select Committee to Study Governmental
>> Operations with Respect to Intelligence Activities</fullname></author>
>>
>> Bruce
>
> -- 
> Markus Hoenicka
> mar...@ca...
> (Spam-protected email: replace the quadrupeds with "mhoenicka")
> http://www.mhoenicka.de
>

RE: [Refdb-users] Installing refdb

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

[please cc the mailing list as this seems to be of general interest]

Hi,

Arthur Chtcherba writes:
 > Hello Marcus,
 > 
 >   Sorry for blunt message, but I think installation process is well beyond
 > the amount of time one would wish to spend on a software such as refdb. It
 > is not some application server which will serve millions of hits a day, that
 > one needs to configure it properly to last neety bitty detail.
 >   Firs of all, I am not a perl programmer and hence was not sure what is
 > organization of perl modules is, so that I could go on and install them. I
 > am running Suse 8.2... So first I did suse update and installed all perl
 > modules it was offering. Unfortunately not all modules required for refdb
 > are standard at least with Suse. Then I went to CPAN and fished out
 > remaining modules, and installed like all perl modules get
 > installed.

There's the refdb-starterkit for exactly this purpose. It contains the
RefDB sources along with all non-standard library sources and the
required Perl modules (except Text::Iconv which I regrettably forgot
to throw in, but I'll fix that asap). While it doesn't auto-install,
it gets you at least all sources you'll need to get things up and
running.

 > Then I
 > had to figure out all these XSL stylesheet paths ... Fortunately these came
 > with Suse...
 > 

Unfortunately there's no really standard way for dealing with SGML and
XML files that would allow to check a handful of default paths and be
sure to find the stylesheets. Every OS and every distro tends to hide
these files somewhere else. It won't work without some user
intelligence.

 >   After doing all that
 > 
 >  refdbd -s -e 0 -l 7
 >  
 > Gives me the following no matter what I try
 > ----------------------------------
 > dbi_driver_dir went to:
 > 
 > dbi is using default deriver dir
 > Unable to initialize libdbi! Make sure you specified a valid driver
 > directory.
 > ------------------------------------
 > 
 > I also installed C dbi library "libdbi.so", but that did not help.
 > 

Word has not spread everywhere that the former libdbi project was
split into libdbi (the framework) and libdbi-drivers (the database
drivers). Apparently you've got only the framework but no database
driver. Look for that package or build it from the sources. When doing
the latter, make sure to specify at least one driver for your database
engine.

 > My question is why spend this effort writing inhalation instructions, etc.
 > and instead have a program which automatically checks for all necessary
 > components, and if not install them, at least directly point where one can
 > download them. Why it was not possible to detect this dbi problem during
 > install, why does it appear during starting the daemon ?

It is fairly easy to test for the presence of libdbi.so as it has to
be in one of the standard library directories. The drivers are loaded
at runtime from either a compiled-in directory or from a directory
that the application specifies. A standard configure test will have a
hard time to cope with this. As for the download hints, the manual has
all the pointers. This is certainly not as convenient as some
auto-installer, but the information is available.

 > 
 >   Why is not possible to have for example some web-based configuration
 > utility.

It is possible for sure but no one wrote a config utility yet. Free
software does not write itself, it's always some warm body
volunteering.

 > > Do you suggest a different installation procedure?
 > 
 > Well I only speak from the user point of view. I find installation procedure
 > which amounts to
 >   "make install"
 > not really sophisticated. All it does just copies files. 

You didn't watch closely enough. "make install" indeed simply shuffles
files because the interesting part happens during "make". About a
dozen files are sent though sed in order to configure stylesheet
paths, config file paths, hostnames, database server prefs, paths to
helper applications etc.

 > It could check for settings, install missing libs,

How would you install missing libs? Build them from sources? Debian
users would be rightfully upset if RefDB bypasses their elegant
package system and messes around with the system. Use .debs? RPM-based
Linuxes won't like them. Ever thought about Cygwin which uses tar.bz2
packages? What about Mac users that have the choice between Fink and
DarwinPorts? Should we offer Free/Net/OpenBSD users a package or a
port? Ever dealt with Solaris? Portability does not come for free,
unfortunately.

 > packages
 >   offer to install a script into "init.d", 

Not all supported systems use the SysV-style init.d stuff to start
daemons. The BSDs (and some Linuxes) use the, well, BSD-style rc.d
stuff, and on Cygwin you want refdbd to run as a native Windows
service.

 >   ask for other configuration settings instead of directing user to edit
 > plain files, 

As noted previously, the config files are preconfigured appropriately
if you use the ./configure parameters wisely. Everything else should
work except the database username/password settings that you'll have
to provide.

 > One can write such app. With minimal effort ...
 > I think... But maybe I am wrong ...
 > 

This sounds very much like an offer. If you want to scratch an itch,
please go ahead. But keep in mind that it won't suffice that such an
installer runs on Suse. A shell script (sh, not bash) is probably the
most advanced program that you can expect to be fully portable.

Besides dealing with these end-user installation hassles, we should
all keep in mind that most operating systems support some packaging
mechanism. A package is way more desirable than an
application-specific installer that will never catch every OS and
distro out there. If your OS does not yet have packages for RefDB and
the required libraries/Perl modules, why don't you try to lobby for
these packages. Try to find someone who creates the required packages
or at least helps you to do it.

regards,
Markus

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

[Refdb-users] corporate names

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

Hi Bruce,

I assume you're talking about risx export (vs. support)?. The content
model in the DTD looks like:

<!ELEMENT author (name|(lastname?,firstname?,middlename*,suffix?))>

so if you write risx datasets you can use the name element for
corporate names and such. However, names are currently handled as
strings internally, so the export routines do not have the faintest
idea that a name once was a corporate name. This is a shortcoming of
the current data model which can be fixed. I assume that MODS has a
similar if not better content model for author names so the effort
wouldn't be wasted.

I've been working on improved support for RIS as understood by
RefMan10. This will require a change of the database schema anyway,
and this may be an appropriate time to fix the name handling as well.

regards,
Markus

Bruce D'Arcus writes:
 > 
 > I just realized something.  The risx support has no concept of 
 > corporate names, which seems a problem to me.  Example:
 > 
 > <author><lastname>U.S. Senate Select Committee to Study Governmental 
 > Operations with Respect to Intelligence Activities</lastname></author>
 > 
 > I realize ris/refman likewise has no such concept, but I wonder if we 
 > can say if there is only one 'name' then the following might be better:
 > 
 > <author><fullname>U.S. Senate Select Committee to Study Governmental 
 > Operations with Respect to Intelligence Activities</fullname></author>
 > 
 > Bruce

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

[Refdb-users] corporate names

[Bruce D'A. <bd...@fa...>]

I just realized something.  The risx support has no concept of 
corporate names, which seems a problem to me.  Example:

<author><lastname>U.S. Senate Select Committee to Study Governmental 
Operations with Respect to Intelligence Activities</lastname></author>

I realize ris/refman likewise has no such concept, but I wonder if we 
can say if there is only one 'name' then the following might be better:

<author><fullname>U.S. Senate Select Committee to Study Governmental 
Operations with Respect to Intelligence Activities</fullname></author>

Bruce

[Refdb-users] Installing refdb

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

Hi Arthur,

I'm afraid I don't fully understand your message. First of all, if
something goes wrong with RefDB, it is very helpful to see the error
messages you run into. If the problem seems related to libdbi, please
include at least the output of refdbd -s -e 0 -l 7 (this should list
the directory where libdbi looks for drivers and the drivers that can
actually be loaded). If you see something strange here, please go back
to the libdbi-drivers package and do a make check. This will show
whether your libdbi and libdbi-drivers installations are functional.

Your last sentence requires a little elucidation: Do you think the
manual is inappropriate? Do you think the installation instructions
should be changed? Do you suggest a different installation procedure?
Please explain and I'll be happy to answer your question.

regards,
Markus

Arthur Chtcherba writes:
 > Unselectively I decided to try out refdb. After 6 hours of installing all
 > necessary perl modules on my linux box, the damn thing still does not work.
 >    Apparently my default dbi path is wrong (it shows up empty).
 > 
 > I feel pretty dumb, after holding highest degree in computer science. I
 > think people here just wasting if not only their own town time, but others
 > too ! why spend so much time writing manuals and just develop normal
 > installation ?
 > 

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

[Refdb-users] Installing refdb

[Arthur C. <ch...@cs...>]

Unselectively I decided to try out refdb. After 6 hours of installing all
necessary perl modules on my linux box, the damn thing still does not work.
   Apparently my default dbi path is wrong (it shows up empty).

I feel pretty dumb, after holding highest degree in computer science. I
think people here just wasting if not only their own town time, but others
too ! why spend so much time writing manuals and just develop normal
installation ?

-- Arthur


 
 
 

Re: [Refdb-users] risx output and encoding?

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

Hi Bruce,

I've added utf-8 as the default encoding for risx output. I've wanted
to do this in the first place but I simply forgot. utf-8 is the most
reasonable default for the XML output, and if anyone needs something
else a simple script using iconv and sed can fix this.

regards,
Markus

Bruce D'Arcus writes:
 > Can't there be a way to specify an encoding default somewhere?  In my 
 > case, it would be utf-8.  The simple solution would thus simply involve 
 > the understanding that I was feeding it utf data, and then the output 
 > declaration would be included.  Ideally you'd want to be able to have 
 > encoding conversion, but that could come later if necessary.
 > 

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

Re: [Refdb-users] ANN: refdb-0.9.4-pre1 available for testing

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

Hi Marc,

I believe this is caused by a subtle typo in the script that screws up
the parsing of the command line parameters. Please try replacing the
following line in runbib:

while getopts ":d:hi:I:N:S:t:u:w:" opt; do

with this one (notice the missing colon after "I"):

while getopts ":d:hi:IN:S:t:u:w:" opt; do

regards,
Markus

Marc Baaden writes:
 > runbib -d mainbib -I -N 8 -S "Art.Rapact.Mbfr." -t db31x collab.xml
 > server error: style spec error0 reference(s) formatted, 0 failed
 > 
 > mv: cannot stat `.dsl': No such file or directory
 > make: *** [collab.bib.xml] Error 1
 > 
 > and it creates two files
 > 8.bib.sgml.err
 > -D.dsl
 > 
 > Note: I use XML, not SGML, strange, isn't it ?
 > My original modified version of runbib still works (for XML).
 > If you want to check against it, look at
 > http://baaden.nerim.net/pub/mybib.csh
 > 

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

[Refdb-users] risx output problems

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

Hi Bruce,

I've added these references to my test database. I think I've fixed
these problems. Please try again with the next prerelease.

regards,
Markus

Bruce D'Arcus writes:
 > 
 > Ok, with the new release, I'm still having screwed up output with two 
 > types of references.  This is just stuf that is invalid against the 
 > DTD; I've not had time to dig deeper.
 > 
 > MPCT
 > CASE

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

Re: [Refdb-users] ANN: refdb-0.9.4-pre1 available for testing

[Marc B. <ba...@sm...>]

Hi Markus,

I finally had some time for testing.

>>> "Markus Hoenicka" said:
 >> I've fixed the worst bugs reported after releasing 0.9.3 and added a
 >> couple of features requested lately. See below for the details. I ask
 >> everyone with a little spare time to test drive this prerelease. The
 >> new archive can be found right here:
 >> [..] 
 >> - added support for multiple middle names in risx.dtd and in refdbd

This works fine now


 >> - added support for arbitrary starting number of numbered
 >>   bibliographies

I could try this using refdbib. The implementation in refdbib is fine,
but I have problems with runbib. See below.


 >> - added a switch to runbib in order to skip ID extraction
Similar, problem with runbib.

Here is the error:
runbib -d mainbib -I -N 8 -S "Art.Rapact.Mbfr." -t db31x collab.xml
server error: style spec error0 reference(s) formatted, 0 failed

mv: cannot stat `.dsl': No such file or directory
make: *** [collab.bib.xml] Error 1

and it creates two files
8.bib.sgml.err
-D.dsl

Note: I use XML, not SGML, strange, isn't it ?
My original modified version of runbib still works (for XML).
If you want to check against it, look at
http://baaden.nerim.net/pub/mybib.csh


Thanks for any pointers or hints,
Marc

-- 
 Dr. Marc Baaden  - Institut de Biologie Physico-Chimique, Paris
 mailto:ba...@sm...      -      http://www.marc-baaden.de
 FAX: +49 697912 39550  -  Tel: +33 15841 5176 ou +33 609 843217