[Refdb-devel] Web Interface

[Sandy Rutherford]

[Markus Hoenicka]

Hi Sandy,

I have been looking for something to manage bibliographical databases
for a project that I manage at IRMACS (http://www.irmacs.ca). refdb
seems to do what I want, but it would help a lot if the web interface
were better.

:-( Unfortunately I have to agree. The current version is a reimplementation of
the previous CGI version and barely makes use of the capabilities of PHP. I
would like to see at least clickable author names and such, plus reasonable
forms to manually enter bibliographic data.

I am willing to invest time and effort into the PHP interface, but
obviously would like to coordinate with current development. Who is
working on the PHP interface?

You're welcome to do so. The PHP version was contributed by Tom Bartolucci. I've
fiddled a bit with the code, mainly to make the installation more portable. But
since then essentially no one has touched the code. The friction caused by
coordination should be close to nil.

If you're serious about a web interface, feel free to start with completely
fresh ideas. There is great stuff out there. Today I came across
http://www.openlaszlo.org/ which could turn your web browser into a EndNote
lookalike. Use whatever is likely to allow you to finish something with the
least amount of effort.

regards,
Markus

[Sandy Rutherford]

Markus,

Post by Markus Hoenicka
Hi Sandy,

I have been looking for something to manage bibliographical databases
for a project that I manage at IRMACS (http://www.irmacs.ca). refdb
seems to do what I want, but it would help a lot if the web interface
were better.

:-( Unfortunately I have to agree. The current version is a reimplementation of
the previous CGI version and barely makes use of the capabilities of PHP. I
would like to see at least clickable author names and such, plus reasonable
forms to manually enter bibliographic data.

Yes, a better form interface for entering data is number 1 on my list.

Post by Markus Hoenicka

I am willing to invest time and effort into the PHP interface, but
obviously would like to coordinate with current development. Who is
working on the PHP interface?

You're welcome to do so. The PHP version was contributed by Tom Bartolucci. I've
fiddled a bit with the code, mainly to make the installation more portable. But
since then essentially no one has touched the code. The friction caused by
coordination should be close to nil.

I'll dive in then and keep you up to date with development.

Post by Markus Hoenicka
If you're serious about a web interface, feel free to start with completely
fresh ideas. There is great stuff out there. Today I came across
http://www.openlaszlo.org/ which could turn your web browser into a EndNote
lookalike. Use whatever is likely to allow you to finish something with the
least amount of effort.

I'll check into openlaszlo. I wasn't aware of it. Of course with all
of these things, one needs to decide if the extra overhead is worth
the gains.

For us the big advantage of refdb is the ability to work with BiBTeX
databases. In the mathematical sciences, TeX is the defacto standard
for publishing.

Sandy

[Sandy Rutherford]

Hi Stéphane,

You can also have a look at refbase (http://refbase.sf.net) which has
another web interface, just to not start from scratch.

How do refdb & refbase compare? I haven't run refbase, but from first
glance the database back-end, and certainly the documentation, for
refdb seem more solid.

Sandy

Stéphane
--
Stéphane Téletchéa, PhD. http://www.steletch.org
Unité Mathématique Informatique et Génome http://migale.jouy.inra.fr/mig
INRA, Domaine de Vilvert Tél : [phone]
78352 Jouy-en-Josas cedex, France Fax : [phone]

[Stéphane Téletchéa]

Post by Sandy Rutherford
Hi Stéphane,

You can also have a look at refbase (http://refbase.sf.net) which has
another web interface, just to not start from scratch.

How do refdb & refbase compare? I haven't run refbase, but from first
glance the database back-end, and certainly the documentation, for
refdb seem more solid.
Sandy

refdb is far more complete (more sql server: sqlite, mysql, postgresql),
more input/output but more complex to implement.

I was providing you the link to refbase since i'm packaging both (for
Mandriva Linux) for my own needs (soon, Mark, too much holidays on May
to just finish the job, but packaging is ok, i just want to the all
before putting on the servers).

Refbase is web-based only (so there are pros and cons), it just another
implementation example (since it is gpl, you can take some code from it).

Yours,

Stéphane

[Sandy Rutherford]

Hi Stéphane & Markus,

Giving this some initial thought, it strikes me that key to having a
nice user interface is to allow users to set db-dependent user
preferences. The RIS specification is too long to imagine having an
unconfigurable user interface, which is efficient. The obvious place
to store the preferences is in refdb itself. The table t_user could
have a third field, user_prefs, which is an alphanumeric string. The
user's preferences could be stored there as PHP code and sucked in and
executed as necessary.

Does this seem reasonable?

Sandy

[Markus Hoenicka]

Hi St=E9phane & Markus,
Giving this some initial thought, it strikes me that key to having a
nice user interface is to allow users to set db-dependent user
preferences. The RIS specification is too long to imagine having an
unconfigurable user interface, which is efficient. The obvious place
to store the preferences is in refdb itself. The table t_user could
have a third field, user_prefs, which is an alphanumeric string. The
user's preferences could be stored there as PHP code and sucked in and
executed as necessary.
Does this seem reasonable?

This certainly would not hurt. Other interfaces might use such a field for
similar purposes, whereas e.g. a command line interface would simply ignore
this field. Implementing this would require two changes:

- add client functions which store and retrieve the preferences

- change the management of the t_user table. Currently an entry in this table is
created whenever a user stores or picks his first reference, and it is deleted
as soon as the user's last reference is deleted or dropped. The entry would
have to be created and deleted through the adduser/deleteuser commands.

regards,
Markus

[Sandy Rutherford]

Hi Markus,

Post by Markus Hoenicka

Post by Sandy Rutherford
user's preferences could be stored there as PHP code and sucked in and
executed as necessary.
Does this seem reasonable?

This certainly would not hurt. Other interfaces might use such a field for
similar purposes, whereas e.g. a command line interface would simply ignore

That sounds reasonable. For other interfaces to use it, best would be
to give it a generic format. Something like:

Key = Val

entries. In that case, it would be the responsibility of the user
interface to find the keys that it needs, parse them, and assign
default values to keys that it can't find. Storing this field as PHP
code would cause grief for other user interfaces. Some of the keys
would likely be user interface specific, but others could be
standardised. This would allow appropriate user prefs to be preserved
across change of user interface. A typical example, would be the
default RIS fields to display when entering a new entry.

Sandy

[Stéphane Téletchéa]

Post by Markus Hoenicka
Hi Sandy,

I have been looking for something to manage bibliographical databases
for a project that I manage at IRMACS (http://www.irmacs.ca). refdb
seems to do what I want, but it would help a lot if the web interface
were better.

:-( Unfortunately I have to agree. The current version is a reimplementation of
the previous CGI version and barely makes use of the capabilities of PHP. I
would like to see at least clickable author names and such, plus reasonable
forms to manually enter bibliographic data.

I am willing to invest time and effort into the PHP interface, but
obviously would like to coordinate with current development. Who is
working on the PHP interface?

You're welcome to do so. The PHP version was contributed by Tom Bartolucci. I've
fiddled a bit with the code, mainly to make the installation more portable. But
since then essentially no one has touched the code. The friction caused by
coordination should be close to nil.
If you're serious about a web interface, feel free to start with completely
fresh ideas. There is great stuff out there. Today I came across
http://www.openlaszlo.org/ which could turn your web browser into a EndNote
lookalike. Use whatever is likely to allow you to finish something with the
least amount of effort.
regards,
Markus

You can also have a look at refbase (http://refbase.sf.net) which has
another web interface, just to not start from scratch.

Stéphane