bibsonomy seems nice: a combination of citeulike and del.icio.us. I'm currently trying it, and will collect a few thoughts and ideas here. The first impression is that the user interface has a nice concept, but it buggy in parts.
- bibtex import
- lets you select a separator char for your keywords (tags); however, when importing keywords separated by ", " (comma space), i end up with tags that have a literal "," at the end. IMHO the import could simply be lenient and split on [\s,;/|]+ (maybe an option is needed for the handling of the underscore "_")
- should trim leading/trailing whitespace from URL (and probably from other stuff too)
- sometimes complains about a missing publication year, even if it is given. Apparently, having a malformed value in some other field causes this misleading message.
- manipulation
- the "pick" command is nice (it puts entries into a temporary basket, for mass-editing) using AJAX
- clicking "edit" on an entry opens an edit field for the tags of the entries, inline. Nice. However, when hitting enter, the whole page is reloaded (no AJAX here); What's worse, it goes back to my home page, instead of keeping my current view/filter. That is very annoying.
- Same goes for delete
- When manually adding an bib record
- Concepts
- It's not yet possible to freely create groups. Too bad. It's not obvious how they would work, either.
- There doesn't seem to be a good way to find other users. A last-fm like "neighbours" feature based on common bookmarks and/or tags would be nice.
- Overall, social interaction isn't too well developed. A messaging facility would be nice.
- Duplicate detection is a nice thing, but only listing them is lame. I want a warning when i post something that's already there. Also, it doesn't seem to work too well...
- I don't like the fact that bibliographic records are copied to personal lists, instead of collaborativly maintaining bibliographic records. But maybe such wiki-style editing wouldn't fit too well into their concept.
- Feeds
- The feed doesn't come with proper HTTP headers: Last-Modified is always 1970-01-01... sucks for caching.
- Import/Export Z39.50/MARC
Ah, btw: I couldn't figure out how to report bugs. And also, the database seems to be down quite frequently. Scares me a bit.
(no comments yet)