Known Bugs with the Bibliographic Search
There are a number of known bugs with the search facilities, and the bibliographies
As regards the `home grown' bibliographies, the major bugs are the following:
The existence of duplicates. Duplicate entries exist
across the different bibliographies (of course), but also within them.
Sometimes these are simply literal duplicates (somehow the same entry
got put in two places). It would be relatively easy to set up software
to deal with these. However, it is rather a rare occurrence. More
commonly, the duplicates are not literally the same, they will differ
e.g. in the order the fields come in, or whether the Authors' first
names are given in full or as initials. There can also be minor
differences in titles, names of journals, publishers (especially where
there are e.g. American and British publishers, and the publishers
have several locations.
Often it happens that an item is entered as a "working paper" when it
first appears, then as a journal article.
The CL/MT bibliography is partly composed of partial bibliographies constructed separately for separate, special purposes, as well as relatively systematic bibliographies of conferences and journals. This obviously leads to duplication, and it is very rare that the duplicates happen to be the same form.
- The accuracy of information is not always good.
Not everyone who constructs an entry remembers to put in the date (or
gets it right), or gets the page numbers correct. Often information
about publisher or place of publication is missing.
- Especially the CL/MT bibliography contains material and
information that noone else could possibly be interested in, e.g. the
fact that a book is in the University library, what its Classmark and
book number there is.
Accents are a general problem: there is no
accepted convention for entering them, and they are often missed out,
or handled wrongly.
Inconsistency in formats and conventions. There are
accepted conventions for entering some kinds of publication in `refer'
format, but these do not cover all cases. For example, do unpublished
PhD dissertations get treated as Books or Reports? There is no
standard way in `refer' for entering an editted volume (as opposed to
the papers in it). There are also differences in format between, e.g.
`refer' and `bib' formats (e.g. the former has all the editors put on
one line -- in one field -- while the latter lists them separately,
As regards the searching (etc) software:
Quite apart from the fact that accents are not entered systematically
in the bibliographies, they cause problems to the searching software.
The bibliographies were designed to be searched with some standard Unix
tools. However, a move to a Window/NT server was required, so I had to write
Perl scripts which more or less emulate the original searching software. So far as I
know, these scripts behave reasonably well. But there may be bugs.
The older bibliographies are held in `refer' format, and translated into other
formats on the fly. This translation process is not perfect. Especially translation
into BibTeX format may be problematic. The treatment of accents and other funny
characters is especially problematic.
The newer bibliographies are held in BibTeX format, and translated into other formats
on the fly. Again, this translation process is not perfect, and the treatment of
accents and other funny characters is especially problematic.
19th August 1995, updated March 2003.