Johnson's STOC/FOCS index issued in 1991 contains a wealth of information on papers originally published through those conferences. Look to it first if you need supplementary information on such a paper. Thanks to Robert Freimer, who took the time to incorporate the relevant geometry citations from that book into this bibliography.
The bibindex and biblook programs now support macro strings (thanks, Jeff), so we can start using them in the biblio. While they save some space, their real advantage is that people can tailor journal and conference abbreviations to their own taste when printing. A set of these macros needs to be designed for the geometry biblio, probably expanding to our current "normal" abbreviations to start, and perhaps eventually offering terse, normal, and full flavours.
To minimize the workload for volunteers, these guidelines eschew redundant precision in entries. So do some journals' house styles. However some researchers prefer their references to always include all available pieces of information like number within volume, day and month, and city. If only bibtex could be told when not to print them, both schools of thought could be accommodated together. A convention like optnumber= might work, if it didn't conflict with emacs bibtex.el's interpretation.
The standard bibtex styles use a sort order of author (von surname given jr), year, and then title. We use author (surname given von jr), title, and then year. Our author ordering is much closer to library rules, and using title before year improves the probability that different appearances of the same paper will be grouped together.
The standard bibtex styles have some fixed ideas about how things can be published. For instance, SIGGRAPH conference papers through 1992 appeared in a proceedings which itself appeared as an issue of their newsletter; attempts to enter such items as @inproceedings would not print completely and elicited a warning on the grounds that a proceedings cannot have both volume and issue numbers (i.e. be published as a journal/newsletter issue). Patashnik's Bibtexing notes suggest a Procrustean solution of adjusting the entry's data to fit the fields that do print. So we had to treat such entries as journal articles, with conference title in a note field:
@article{awg-psg-78
, author = "P. Atherton and K. Weiler and D. P. Greenberg"
, title = "Polygon shadow generation"
, journal = "Comput. Graph."
, volume = "12"
, number = "3"
, year = "1978"
, pages = "275--281"
, note = "Proc. SIGGRAPH '78"
}
From 1993 on the SIGGRAPH proceedings is a series separate from the
newsletter and entries are @inproceedings with a booktitle. Perhaps
bibtex might better use a general inclusion field ``in = "siggraph78"''
rather than crossref with its hardwired list of permitted situations,
and in any case try harder to make sensible output from the fields
it finds.
Classic bibtex comes with a limit of 750 entries per bibliography, which is far too small for us. If you want to produce your own hardcopy of significant subsets, you probably need to reconfigure your local bibtex and tex programs. Change files to enlarge the critical parameters in bibtex.web and tex.web are available upon request. (These and other ingredients of the current formatting really ought to be cleaned up and released some day.) Simple parameter fiddling breaks down beyond around 4500 entries when the number of field strings overflows web's 32767 limit on simple macro values. It may be possible to change web someday, but meanwhile a workaround is to tell web some low numbers, then patch the generated C code. You can avoid such issues entirely by using the default Postscript softcopy available from the ftp directory.
We should remember that bibtex is not being used by everyone, and researchers still using refer would like some means of back-converting from the current format. A makeshift program bibtex2refer is included in the distribution for anyone wishing to use it. If you do change or improve it please let us know so that we can distribute the improvements too.
A program "bibview" which is an X tool for manipulating bibtex databases has been published in comp.sources.x/v18i099. Its README says it "supports the user in making new entries, searching for entries and moving entries from one bib to another. It is possible to work with more than one bib simultaneously. bibview is implemented with Xt and Athena Widgets." bibview is available for ftp from (among other places) <ftp://ftp.informatik.tu-muenchen.de:/pub/comp/typesetting/tex/bibview-2.0.tar.Z> and comp.sources.x archive sites. Check archie for a location near you.
The Math Reviews index can be searched online by telnet to e-math.ams.org, login e-math, password e-math. Search on author name returns MR numbers only.
Finally, what can/should be done to improve the keyword system? Its use at present is de facto optional. People seem to think the list has too many categories to keywordize quickly, but too few to keywordize well. Yet such a vitamin supplement to an entry's indexing is invaluable for searching purposes -- it bridges terminology between the author and reader. Consider the following quotation:
"... mathematics is the art of giving the same name to different things .... When the language is well chosen, we are astonished to learn that all the proofs made for a certain object apply immediately to many new objects; there is nothing to change, not even the words, since the names have become the same."Is there some way in which we can get that supplementation without unfairly complicating things for the entering bibliographer? One idea is to boil the system down to a smaller (and more easily maintained) key *vocabulary* which we would then be free to twist and combine as appropriate for a given entry. (We would in effect be defining a keywordspeak language, according to Orwell.) This would take a fair amount of time and expertise to design well.- H. Poincare, Science and Method