refdb-users — List for RefDB users

Re: [Refdb-users] Cannot add reference.

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

Wei-Wei Guo writes:
 > Do you mean patches for refdb-0.9.9-1.tar.gz? I can do that.

Unfortunately I can't reply to the proper message as I didn't receive
your last reply in my inbox. I can see it in the list archives though.

I've checked your log output. I've designed another small patch to
drill a bit deeper. Please apply it in addition to the previous patch
and rebuild RefDB again. The procedure you used last time was
apparently successful.

BTW which type of processor do you use? Is it a non-x86, by any
chance?

regards,
Markus

Re: [Refdb-users] Cannot add reference.

[Wei-Wei G. <ww...@gm...>]

Markus Hoenicka 写道:
> Wei-Wei Guo writes:
>  > Do you mean patches for refdb-0.9.9-1.tar.gz? I can do that.
>  > 
> 
> I've appended a simple patch for one of the source files to provide
> additional debug output. Please apply the patch to the sources with a
> command like:
> 
> patch < refdbdref.c.patch
> 

The patch cannot be applied, so I open refdbdref.c and changed it by hand.

$ patch < refdbdref.c.patch
patching file refdbdref.c
Hunk #1 succeeded at 2114 (offset 1 line).
Hunk #2 FAILED at 2142.
1 out of 2 hunks FAILED -- saving rejects to file refdbdref.c.rej


> from within the src subdirectory of the source tree. Then please
> rebuild this directory and install it, or (whichever is easier for
> you), run the server as a regular app from the src directory, like
> this:
> 
> ./refdbd -s [other options]
> 

Cause I don't know how make install will affect the system consistence,
I just repack the refdb-0.9.9-1.tar.gz with the modified refdbdref.c and
rebuild it by Gentoo emerge. And I run refdb by "/etc/init.d refdb start".
Is what I have done ok?


> Then repeat your "add references" test from another xterm, and send
> the refdbd log. Just ask if anything is unclear, I'll be happy to
> help.

I repeat the process I did before. The refdbd.log is gotten by remove the
old one first and run the test second. It's verbose than the ones get in
previous tests. See the attachment.

BTW. I'm moving flat in those days, so my feedback might be a little late.
Sorry for that.


Best wishes,
Wei-Wei

Re: [Refdb-users] Cannot add reference.

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

Wei-Wei Guo writes:
 > Do you mean patches for refdb-0.9.9-1.tar.gz? I can do that.
 > 

I've appended a simple patch for one of the source files to provide
additional debug output. Please apply the patch to the sources with a
command like:

patch < refdbdref.c.patch

from within the src subdirectory of the source tree. Then please
rebuild this directory and install it, or (whichever is easier for
you), run the server as a regular app from the src directory, like
this:

./refdbd -s [other options]

(kill it with Ctrl-c when done)

Then repeat your "add references" test from another xterm, and send
the refdbd log. Just ask if anything is unclear, I'll be happy to
help.

I've just tested this on a FreeBSD box using the current svn revision
and the sqlite3 driver. Things work ok for me.

regards,
Markus

Re: [Refdb-users] RISX vs. RIS

[Torsten B. <br...@ph...>]

Hallöchen!

Torsten Bronger writes:

> [...]
>
> The data model of my front-end is RISX.  Where can I get
> information about how to interpret the RISX fields for the
> different types?  [...]  But this makes is necessary to map RISX
> to RIS.  Is there information about this?

I investigated further and now I'm slightly confused.  The RefDB
manual says that RISX should be prefered for communicating with the
database, and even that it should be used for backup dumps.
However, it can't hold all the data in the database.  For example,
if I store a book reference, there is a "volume" and an "edition"
field, however in RISX, only "volume" survives.  In other words, the
RISX dump is not lossless.  An RIS dump contains all fields, though.

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...

[Refdb-users] RISX vs. RIS

[Torsten B. <br...@ph...>]

Hallöchen!

I'm now in the phase of choosing labels for the form field in my Web
front-end.  My plan is to use JavaScript to change the label names
according to the reference type.  For example, "publication title"
will be "journal name" for articles and "book title" for book
chapters.

The data model of my front-end is RISX.  Where can I get information
about how to interpret the RISX fields for the different types?
http://refdb.sourceforge.net/risx/index.html is of some help but it
is rather limited.  The most comprehensive tables can be found in
<http://www.referencemanager.com/support/docs/ReferenceManager11.pdf>,
pp 525 ff.  But this makes is necessary to map RISX to RIS.  Is
there information about this?

By the way, RIS seems to have four titles (primary, secondary,
series, periodical), whereas RISX has only three of them.  In case
of conferences, RIS uses all four (see page 527).  How is this
mapped to RISX?

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...

Re: [Refdb-users] Cannot add reference.

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

Hi,

Quoting Wei-Wei Guo <ww...@gm...>:

>> Would you be able to use to test the current svn revision?
>
> I'm using Gentoo and usually install packages by ebuilds provided by others.
> I tried to make a ebuild for svn version. It complains "no configure
> script" when
> compiling. After compare files in svn trunk and refdb-0.9.9-1.tar.gz, I found
> there is no "configure" file in svn trunk. I'm not familiar with this kind of
> things, so I'm not sure whether it's a problem of my ebuild or svn source.
>

No, it's just a matter of running "./autogen.sh" in the top-level  
directory of your svn checkout. This script generates all required  
files for the autotools, including the configure script.

In fact, you don't need to *install* the svn version in order to test  
it. IIRC the current svn revision does not depend on stuff which is  
not installed anyway if you use a 0.9.9-1 ebuild. You could just build  
the svn revision and run that from the build directory, without even  
installing it.

>
>> Would you be comfortable with testing source code patches (to get additional
>> debug output) which I send to you?
>
> Do you mean patches for refdb-0.9.9-1.tar.gz? I can do that.
>

I guess that'll work. I'm using the svn revision, but the code in  
question didn't change that much, so patches against my version are  
likely to apply cleanly to your version.

>
>> And, if nothing else helps, would you be able to provide temporary access
>> via ssh to your box?
>
> I'm not sure. I rarely use ssh. If it's necessary, I can ask someone  
>  teach me.

That's not a problem at this time. I'd like to retrieve as much  
information as possible using some patches before trying anything else.
>
>
> And I did a clear try with a newly created refdbd.log. I send the
> complete log to
> you. Hope it helpful.
>

Thanks for the log. Unfortunately it just confirms what we already  
know: the first dataset is added ok, all subsequent datasets fail  
while refdbd tries to add them to your personal reference list.

regards,
Markus

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

Re: [Refdb-users] Cannot add reference.

[Wei-Wei G. <ww...@gm...>]

Markus Hoenicka 写道:

Hi Markus,

> First of all, which RefDB version are you using? 

I'm using refdb-0.9.9-1.

> Would you be able to use to test the current svn revision? 

I'm using Gentoo and usually install packages by ebuilds provided by others.
I tried to make a ebuild for svn version. It complains "no configure script" when
compiling. After compare files in svn trunk and refdb-0.9.9-1.tar.gz, I found
there is no "configure" file in svn trunk. I'm not familiar with this kind of
things, so I'm not sure whether it's a problem of my ebuild or svn source.


 > Would you be comfortable with testing source code patches (to get additional
 > debug output) which I send to you?

Do you mean patches for refdb-0.9.9-1.tar.gz? I can do that.


> And, if nothing else helps, would you be able to provide temporary access 
 > via ssh to your box?

I'm not sure. I rarely use ssh. If it's necessary, I can ask someone teach me.


And I did a clear try with a newly created refdbd.log. I send the complete log to
you. Hope it helpful.

     $ rm studio/science/biblio/wwguodb
     $ refdba -C createdb wwguodb
     999:1 created:0 failed
     $ refdbc -C addref -A ris testrefs.utf8.ris
     999:1 added:0 skipped:21 failed
     $ cp /var/log/refdbd.log .


Best wishes,
Wei-Wei

Re: [Refdb-users] Cannot add reference.

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

Hi,

Quoting Wei-Wei Guo <ww...@gm...>:

> Since I don't get your feedback, I'm wandering whether you get it. So I
> send it again. If you get duplicated email, sorry for that.
>

I'm sorry but I didn't receive those mails. For whatever reasons, the  
mail you sent to the list went through ok.

>     $ refdbc -C addref -A ris testrefs.utf8.ris
>     999:1 added:0 skipped:21 failed
>
>     $ sqlite3 studio/science/biblio/wwguodb
>     SQLite version 3.6.10
>     sqlite> SELECT note_id, note_user_id FROM t_note WHERE   
> note_key='wwguo-wwguo';
>     1|1
>

Unfortunately this doesn't make sense. The only reasonable way for  
RefDB to return the error which you originally posted is if the above  
SELECT fails, but it returns the expected values. Strange.

Now, what I'll do is to upgrade my SQLite (I think I run something  
older than 3.6.10) and see if I can reproduce the problem. If I don't  
succeed, we'd have to analyze more closely what's happening on your box.

First of all, which RefDB version are you using? Would you be able to  
use to test the current svn revision? Would you be comfortable with  
testing source code patches (to get additional debug output) which I  
send to you? And, if nothing else helps, would you be able to provide  
temporary access via ssh to your box?

regards,
Markus


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

Re: [Refdb-users] Cannot add reference.

[Wei-Wei G. <ww...@gm...>]

Hi Markus,

Sorry I didn't notice I keep on sending my replies to your email instead of
to refdb-users.

Since I don't get your feedback, I'm wandering whether you get it. So I
send it again. If you get duplicated email, sorry for that.


Markus Hoenicka 写道:
> Quoting Wei-Wei Guo <ww...@gm...>:
> 
>>     7:pid=3326:Sun Apr 26 16:51:49 2009:SELECT note_id, note_user_id  
>>  FROM t_note WHERE note_key='wwguo-wwguo'
> 
> This appears to be the command which fails, or which causes a failure  
> a few lines further down. Could you please run this command in sqlite3  
> using one of your test databases after you tried to add references?
> 

    $ rm studio/science/biblio/wwguodb

    $ refdba -C createdb wwguodb
    999:1 created:0 failed

    $ refdbc -C addref -A ris testrefs.utf8.ris
    999:1 added:0 skipped:21 failed

    $ sqlite3 studio/science/biblio/wwguodb
    SQLite version 3.6.10
    sqlite> SELECT note_id, note_user_id FROM t_note WHERE note_key='wwguo-wwguo';
    1|1


Best wishes,
Wei-Wei

Re: [Refdb-users] Cannot add reference.

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

Quoting Wei-Wei Guo <ww...@gm...>:

>     7:pid=3326:Sun Apr 26 16:51:49 2009:SELECT note_id, note_user_id  
>  FROM t_note WHERE note_key='wwguo-wwguo'

This appears to be the command which fails, or which causes a failure  
a few lines further down. Could you please run this command in sqlite3  
using one of your test databases after you tried to add references?

thanks
Markus

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

Re: [Refdb-users] Cannot add reference.

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

Hi,

Quoting Wei-Wei Guo <ww...@gm...>:

>     4:pid=3326:Sun Apr 26 16:51:49 2009:inserting reference data failed

thank you for posting most of the relevant information along with your  
problem. The only thing I'd like to know in addition is the type of  
reference data which you try to add. The error above is often caused  
by trying to add data without using the proper -A switch. If refdbd  
attempts to read risx data as ris or vice versa, it'll fail right away.

If you're not sure, please send the exact command which you issue to  
add the data, along with the dataset(s).

regards,
Markus


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

[Refdb-users] Cannot add reference.

[Wei-Wei G. <ww...@gm...>]

Dear all,

I'm stuck in the first step when trying to get my RefDB working.
I cannot add reference.

I doesn't meet any error when run the command 'refdbc'.

Here is part of the log info. Every entry has a part of this.

    7:pid=3326:Sun Apr 26 16:51:49 2009:BEGIN
    7:pid=3326:Sun Apr 26 16:51:49 2009:SELECT refdb_id, refdb_citekey FROM t_refdb WHERE refdb_id=6
    6:pid=3326:Sun Apr 26 16:51:49 2009:ignore numerical ID
    7:pid=3326:Sun Apr 26 16:51:49 2009:INSERT INTO t_refdb (refdb_type,refdb_citekey) VALUES ('DUMMY','DUMMYwhiz3326')
    7:pid=3326:Sun Apr 26 16:51:49 2009:SELECT user_id FROM t_user WHERE user_name='wwguo'
    7:pid=3326:Sun Apr 26 16:51:49 2009:INSERT INTO t_xuser (user_id, refdb_id) VALUES (1,3)
    7:pid=3326:Sun Apr 26 16:51:49 2009:SELECT refdb_id, refdb_citekey FROM t_refdb WHERE refdb_id=3
    7:pid=3326:Sun Apr 26 16:51:49 2009:SELECT note_id, note_user_id FROM t_note WHERE note_key='wwguo-wwguo'
    4:pid=3326:Sun Apr 26 16:51:49 2009:inserting reference data failed
    7:pid=3326:Sun Apr 26 16:51:49 2009:ROLLBACK
    4:pid=3326:Sun Apr 26 16:51:49 2009:failed processing dataset

I tried the database with

    $ sqlite3 /home/wwguo/biblio/wwguodb
    sqlite> INSERT INTO t_refdb (refdb_type,refdb_citekey) VALUES ('DUMMY','DUMMYwhiz3326');

And it's works.

Here is my 'refdbdrc':

    # refdbdrc
    refdblib /usr/share/refdb
    main_db	refdb
    serverip	localhost
    dbsport		
    dbserver	sqlite3
    dbpath		/home/wwguo/studio/science/biblio
    port	9734
    logfile /var/log/refdbd.log
    logdest file
    loglevel debug
    pidfile		/var/run/refdbd.pid
    remoteadmin f
    remoteconnect f
    keep_pnames t
    keyword_scan t
    db_encoding	UTF-8
    in_encoding	UTF-8

My 'refdbcrc':

    # refdbcrc
    serverip	127.0.0.1
    port	9734
    pager	less
    username	wwguo
    passwd
    defaultdb	wwguodb
    logdest file
    loglevel info
    logfile /var/log/refdbc.log
    cssurl /usr/share/refdb/css/refdb.css
    fromencoding UTF-8
    toencoding UTF-8


Please give some help! Any suggestion is appreciated.

Best wishes,
Wei-Wei

Re: [Refdb-users] RefDB and privacy

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

Quoting Torsten Bronger <br...@ph...>:

> But it is not only <link>, it is also <notes>.  Is <notes> realised
> with extended notes?
>

No, <notes> are stored in a simple text field, together with the  
reprint status (i.e. separately for each user). Extended notes are  
entirely separate.

regards,
Markus

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

Re: [Refdb-users] RefDB and privacy

[Torsten B. <br...@ph...>]

Hallöchen!

Markus Hoenicka writes:

> Zitat von Torsten Bronger <br...@ph...>:
>
>> This reminds me that I wanted to ask about the privacy policy in
>> RefDB.  From my tests it seems that I can see all <libinfo>
>> datasets of all other users.  Is this correct, or how can I
>> change this?
>
> I think this is correct. The idea was to make the location of
> other copies of an article available if you don't have a copy of
> your own. I have no idea whether this works in a real-world
> setting. I was also under the impression that there would be no
> demand to hide offprints from co-workers, so in contrast to
> extended notes (which can be kept private) links are supposed to
> be public.

But it is not only <link>, it is also <notes>.  Is <notes> realised
with extended notes?

> Would you prefer to treat link as private data, maybe optionally?

Actually I like liberal settings.  Especially because I think RefDB
is used in rather small groups < 100 people without anonymity where
excessive permission management is not vitally important.

Refbase plans to add fine-grained permission control, but I think
this is overkill.  It would be nice however if RefDB's log facility
enabled the admin to detect vandalism.

In Django-RefDB, I will install only one permission, namely the
permission to edit entries other than yours.  Django-RefDB keeps
track who created an entry, and you can always edit your entries,
but not those of others, unless you have that permission.

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...

Re: [Refdb-users] RefDB and privacy (was: About the AV and L1 fields)

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

Zitat von Torsten Bronger <br...@ph...>:

> This reminds me that I wanted to ask about the privacy policy in
> RefDB.  From my tests it seems that I can see all <libinfo> datasets
> of all other users.  Is this correct, or how can I change this?
>

I think this is correct. The idea was to make the location of other  
copies of an article available if you don't have a copy of your own. I  
have no idea whether this works in a real-world setting. I was also  
under the impression that there would be no demand to hide offprints  
from co-workers, so in contrast to extended notes (which can be kept  
private) links are supposed to be public. Would you prefer to treat  
link as private data, maybe optionally?

regards,
Markus

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

Re: [Refdb-users] How to mark institute's publication

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

Zitat von Bruce D'Arcus <bda...@gm...>:

> Aren't they all hacks of sort that abuse the (limited) RIS semantics?
> What you're really wanting to model is that a contributor is
> affiliated with a particular institution (your's), and then to grab
> items based on that affiliation.
>

The problem with this approach in life sciences is that most original  
datasets (e.g. Pubmed) record only the affiliation of the first  
author. You would miss all items that originate from somewhere else  
but are co-authored by members of your institute not because RIS  
semantics are limited, but because these data are usually not  
available. You'd have to manually add affiliations for co-authors  
which is about as cumbersome as using one of the suggested hacks. BTW  
I'd favour your (Bruce) suggestion to add a custom keyword.

regards,
Markus


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

[Refdb-users] RefDB and privacy (was: About the AV and L1 fields)

[Torsten B. <br...@ph...>]

Hallöchen!

Markus Hoenicka writes:

> [...]
>
> [...] However, there is one thing to keep in mind. The current
> implementation uses a simple check for "file://" at the start of
> each link to make sure a libinfo link is an URL pointing to a file
> resource owned by the user, and not a public URL accessible via
> http or ftp. That is, libinfo links using any protocol other than
> "file://" or none at all end up being added as public links. This
> code has been added too long ago to exactly remember why it was
> added. It now looks a little strange to me, and I'm open to
> suggestions whether or not to remove this restriction.

This reminds me that I wanted to ask about the privacy policy in
RefDB.  From my tests it seems that I can see all <libinfo> datasets
of all other users.  Is this correct, or how can I change this?

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...

Re: [Refdb-users] About the AV and L1 fields

[Torsten B. <br...@ph...>]

Hallöchen!

Markus Hoenicka writes:

> Torsten Bronger writes:
>
>> Are pubinfo and libinfo inextricably intertwined or perfectly
>> separated?
>
> They are supposed to be treated separately. Your example helped to
> fix two unrelated bugs which caused the behaviour you saw.

Thank you!

> [...]
>
> [...] However, there is one thing to keep in mind. The current
> implementation uses a simple check for "file://" at the start of
> each link to make sure a libinfo link is an URL pointing to a file
> resource owned by the user, and not a public URL accessible via
> http or ftp. That is, libinfo links using any protocol other than
> "file://" or none at all end up being added as public links. This
> code has been added too long ago to exactly remember why it was
> added. It now looks a little strange to me, and I'm open to
> suggestions whether or not to remove this restriction.

In my opinion, it should be removed.  If a private link is turned
into a public one implicitly, this is a source of irritation.

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...

Re: [Refdb-users] How to mark institute's publication

[Torsten B. <br...@ph...>]

Hallöchen!

Bruce D'Arcus writes:

> On Sat, Apr 18, 2009 at 6:51 AM, Torsten Bronger
> <br...@ph...> wrote:
>
>> Okay, I see four possibilities:
>>
>> * Use a U1..U5 field anyway
>>
>> * Add a bogus series (aka set) with the title "IEF-5 publications"
>>
>> * Use RISX' <address> tag with a special micro syntax,
>>  e.g. "[institution: IEF-5]"
>>
>> * Store it in the Web client instead of in RefDB
>
> Aren't they all hacks of sort that abuse the (limited) RIS
> semantics?

RIS is really limited.  Some bibliography standards are monstrous,
but RIS is too simple.  Additionally, it is poorly specified.

On the other hand, RISX is much better, and this is the actual data
model of RefDB.  I think it's a good compromise with only a few
inconveniences.

> What you're really wanting to model is that a contributor is
> affiliated with a particular institution (your's), and then to
> grab items based on that affiliation.

Probably.  However, RISX has no tag for it, so I use the <address>
tag currently.  It even seems to be within its semantics, like in

<address>t.b...@fz...; IEF-5, Jülich Forschungszentrum,
Germany</address>

I just have to make sure that the string "IEF-5, Jülich
Forschungszentrum, Germany" is always *exactly* this.  But since
it's only a checkbox in the Web frontend, my own code can assure
this.

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...

Re: [Refdb-users] How to mark institute's publication

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

On Sat, Apr 18, 2009 at 6:51 AM, Torsten Bronger
<br...@ph...> wrote:

> Okay, I see four possibilities:
>
> * Use a U1..U5 field anyway
>
> * Add a bogus series (aka set) with the title "IEF-5 publications"
>
> * Use RISX' <address> tag with a special micro syntax,
>  e.g. "[institution: IEF-5]"
>
> * Store it in the Web client instead of in RefDB

Aren't they all hacks of sort that abuse the (limited) RIS semantics?
What you're really wanting to model is that a contributor is
affiliated with a particular institution (your's), and then to grab
items based on that affiliation.

In the Drupal-based biblio module, one can associate a a contributor
with a Drupal user, and then use that to generate reports and such. So
I guess his approach was your option #4.

There's a fifth possibility, BTW: use a keyword/tag.

Bruce

[Refdb-users] How to mark institute's publication

[Torsten B. <br...@ph...>]

Hallöchen!

For a change, I have a question which may also be answered by
readers other than Markus.  ;-)

I need to mark publications which were made by our scientific
institute, in order to be able to automate the generation of the
publication report which needs to be submitted annually in our
research centre.

I try to avoid the U1-U5 fields of RIS because I don't like their
arbitrary semantics, and I run out of them quickly.

Okay, I see four possibilities:

* Use a U1..U5 field anyway

* Add a bogus series (aka set) with the title "IEF-5 publications"

* Use RISX' <address> tag with a special micro syntax,
  e.g. "[institution: IEF-5]"

* Store it in the Web client instead of in RefDB

What do you think?

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...

Re: [Refdb-users] About the AV and L1 fields

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

Torsten Bronger writes:
 > Are pubinfo and libinfo inextricably intertwined or perfectly
 > separated?
 > 

They are supposed to be treated separately. Your example helped to fix
two unrelated bugs which caused the behaviour you saw.

First, a bug prevented public links to be added to the returned string
if pubinfo does not contain anything else but link elements. That is,
the data were correctly added to the database and correctly retrieved,
but they were simply not added to the result string.

Second, a SQL query bug requested both public and private links when
retrieving the libinfo link elements.

This is now fixed in svn. However, there is one thing to keep in
mind. The current implementation uses a simple check for "file://" at
the start of each link to make sure a libinfo link is an URL pointing
to a file resource owned by the user, and not a public URL accessible
via http or ftp. That is, libinfo links using any protocol other than
"file://" or none at all end up being added as public links. This code
has been added too long ago to exactly remember why it was added. It
now looks a little strange to me, and I'm open to suggestions whether
or not to remove this restriction.

regards,
Markus

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

Re: [Refdb-users] About the AV and L1 fields

[Torsten B. <br...@ph...>]

Hallöchen!

Markus Hoenicka writes:

> Torsten Bronger writes:
>
>> But is L2 a personal or a global field?  According to the manual,
>> they "hold information which is stored for each user separately".
>> 
>> The RISX DTD has <link> tags in both the pubinfo (global) and the
>> libinfo (personal) elements.
>
> The sources say that L1 through L4 elements from risx pubinfo elements
> are imported without being attached to a particular user. In contrast,
> L1 through L4 elements from libinfo elements are owned by the
> corresponding user. When you retrieve risx data, you'll get back both
> types, i.e. the "global" pubinfo links and the personal libinfo
> links.

Are pubinfo and libinfo inextricably intertwined or perfectly
separated?

An example which confuses me: I send the following to RefDB with the
updateref command:

    <entry citekey="Bld2006" id="1" type="JOUR">
      <part>
        <title>Electron Spin Resonance in thin film silicon after low
temperature electron irradiation</title>
        <author>
          <lastname>Bl&#246;d</lastname>
          <firstname>H.</firstname>
          <middlename />
        </author>
      </part>
      <publication>
        <title type="full">Thin Solid Films</title>
        <pubinfo>
          <link type="pdf">PUBLIC</link>
        </pubinfo>
      </publication>
      <libinfo user="drefdbuser1">
        <reprint status="INFILE" />
        <link type="pdf">PRIVATE</link>
      </libinfo>
    </entry>


With getref, I get RISX data which contains this:

    <publication>
      <title type="full">Thin Solid Films</title>
      <pubinfo>
        <link type="pdf">PRIVATE</link>
        <link type="pdf">PUBLIC</link>
      </pubinfo>
    </publication>
    <libinfo user="drefdbuser1">
      <reprint status="INFILE"/>
      <link type="pdf">PRIVATE</link>
      <link type="pdf">PUBLIC</link>
    </libinfo>

Actually, although the DTD allows that, I'm quite suprised that two
"pdf" links are possible.  Additionally, why do I find both links in
pubinfo as well as in libinfo?  Can one avoid this?

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...

Re: [Refdb-users] About IDs and citation keys

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

Quoting Torsten Bronger <br...@ph...>:

> This all sounds like the CK in RefDB is what the ID is in general
> database applications: A primary key that doesn't change.
>

Basically yes. Providing both a unique ID and a unique citation key  
combines the best of both worlds: You can file paper copies by ID, and  
electronic offprints by citation key. Without legacy support for paper  
copies, CK would be sufficient.

> My scenario is this: In my RefDB web frontend, the user can upload
> PDF files which are stored in the file system.  Every reference gets
> a sub-directory with its PDF file.  The question is, which name
> should the sub-directory have?  In order to avoid re-naming it, it
> should never change.  At the moment, I use the ID but probably the
> CK is better?
>

I suggest to use CK instead.

regards,
Markus

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

Re: [Refdb-users] About IDs and citation keys

[Torsten B. <br...@ph...>]

Hallöchen!

Markus Hoenicka writes:

> Quoting Torsten Bronger <br...@ph...>:
>
>> I need something that *never* changes for a reference entry, so I
>> plan to use the ID.  My problem is that I don't easily get it
>> after addition of the reference.  I only get the generated
>> citation key.  But the citation key can change.
>
> The citation key is not supposed to change. In fact, it takes
> quite some effort to change the citation key. I'm not aware of a
> simple way using RIS, and risx prefers the citation key over the
> ID if both are specified. Also, if you have to re-create a
> database from a RIS or risx dump, the ID is more likely to change
> than the citation key.

This all sounds like the CK in RefDB is what the ID is in general
database applications: A primary key that doesn't change.

My scenario is this: In my RefDB web frontend, the user can upload
PDF files which are stored in the file system.  Every reference gets
a sub-directory with its PDF file.  The question is, which name
should the sub-directory have?  In order to avoid re-naming it, it
should never change.  At the moment, I use the ID but probably the
CK is better?

> Do you have some scenario in mind which is likely to alter the
> citation key?

No.  I just thought that it can be done by just giving another CK in
the RISX passed to updateref.  But apparently, this doesn't work.

Tschö,
Torsten.

-- 
Torsten Bronger, aquisgrana, europa vetus
                   Jabber ID: tor...@ja...