BibDesk is already a very interesting piece of software and its=20
becoming opensource is a blessing.
Now, I must confess that I'm not a developer. I can read code to some=20
extent and hack a few things, but that's about it. OTOH, I have a lot=20
of ideas about reference management and I see a great future for=20
solutions including BibDesk as a central component.
In fact, BibDesk could very likely do what most people use commercial=20
reference management software for. As it stands, I still use EndNote=20
for basic reference management, sending BibTeX output to BibDesk for=20
citation support in TeXShop.
Here are, in my mind, the necessary features to make EndNote=20
1. Robust transfer filters (import/export)
2. Option for automatic labels
3. Custom bib styles
4. Z39.50 support
1. I know BibDesk is meant to be used with BibTeX files but as it=20
already does some basic file hacking, it probably wouldn't be too hard=20=
to implement a feature similar to "import filters" (and export "output=20=
styles") in Endnote. BibDesk could keep .bib format as the internal=20
representation but importing and exporting files from common formats=20
would make BibDesk extremely useful. An XML format (with accompanying=20=
XSLT and such) would ease standardization.
2. As BibTeX uses labels, importing references also means inputting=20
labels for each reference. This can be done manually for those who=20
really want to control them but automatic labels (based on author and=20
date, for instance) would be much more efficient in automatic imports.
3. BibTeX is awfully nice but .bst's are difficult to manage and=20
customize. If BibDesk were to integrate a feature simplifying this,=20
there could be a wealth of custom formats for almost any publication.
4. Directly importing references from library catalogs and online=20
databases is, in my mind, the best possible feature a reference=20
management program can have. It's in fact the main reason I still use=20
EndNote. While such a feature looks complex, there seems to be=20
framework for implementing Z39.50 functionalities.
Anyway, I just wanted to generate some discussion. I do believe some of=20=
these features could be implemented without too much effort.
Ph.D. Candidate, Department of Folklore and Ethnomusicology, Indiana=20
Lecturer, D=E9partement d'anthropologie, Universit=E9 de Montr=E9al