Print

Print


I am a programmer with nearly five years of experience
developing applications full-time in support of finding
aids (first with findaid.dtd and then ead.dtd). I have
also had the opportunity to work with a diverse set of
finding aids from many, many different institutions, not
just those created here at Berkeley. My experience has been
that the different conflicting ways in which ead is applied
has made development of generic tools and stylesheets com-
pletely impossible. Matthew Nickerson's recent posting to
the ead list on this topic really hit a nerve with me. One
of the first things that this project should have been able
to provide was a pool of stylesheets and application tools
that everyone could share. Instead, those institutions which
have the resources to develop their own tools have done so
and everybody else has been left out in the cold.

I have been able to accomodate this conflicting markup
style everywhere else in the finding aid (albeit with
great difficulty) except for the container lists. Most of
my observations have to do with uniform encoding practices
and policies and I will save these comments for the working
group responsible for developing these, when the time comes.
I will take advantage of Kris Kieslings invitation to suggest
two changes to the DTD that would go a long way to solving
this problem.

1. A minor addition would be to add new NOTATIONs for HTML
   and EMAIL to the eadnotat.ent. Currently various institutions
   are accomodating this with their own flavors of NOTATION
   declarations in the DTD subset. These minor variations are
   problematic!

2. A much more useful change would be the addition of a "style"
   attribute to the <dsc> element. My first thoughts are that
   style would not be a closed list but would be numeric. One
   of the responsibilities of an EAD working group would be to
   publish formal specifications for which types of container
   lists would conform to which style. Example:

   style=0: Tabular markup, using <drow>s, <dentry>s, etc.

   style=1: Non-tabular markup. One <unitloc> (or <unitid>)
            followed by one <unittitle> and other information
            such as <scopecontent>, <odd>, etc.

   style=2: Two <unitloc>s followed by a <unittitle> and other
            information.

   style=3: Three <unitloc>s followed by a <unittitle> and other
            information>

       etc.

   Since the format of a container list can change from series
   to series, the <c> tag should also have a style attribute which
   could override the one in the <dsc>. altrender could be used for
   this purpose (and currently is in some institutions!) but I think
   it more useful to formalize something specifically for this use.
   The style attribute of course would not be required.

   An alternative to a numeric style attribute might be an attribute
   of type ENTITY. In a more sophisticated technological environment
   the SGML application could fetch a style description file from a
   server out on the net.

   In either case the important point is that a master list of container
   list types is maintained by an SAA working group which publishes
   the detailed markup specifications.

   This attribute would be a very slight intrusion of formatting
   information into the intellectual purity of a finding aid (We
   already have this intrusion in a big way with tabular markup of
   course. The style attribute might eliminate the need for tabular
   markup almost entirely). But the tremendous usefulness it would
   add to the way in which we display and otherwise work with our
   finding aids makes it well worth the addition of one, tiny attribute.
   We could then create pools of stylesheets. A finding aid author knows
   which style of container list she has and could download the
   appropriate stylesheet from a central web site. The attribute
   would greatly facilitate the interchange of finding aids between
   institutions and make the establishment of a union database of
   finding aids far easier.

On another note, I have heard it often said that every <c> should
have exactly one <unittitle>. This makes a great deal of sense to
me but I have never seen it formally stated, either in the DTD, the
tag library, or the application guidelines. I'm not sure it's ever
been said on the ead list. If this is the case, can we formalize
this "law" somehow? It's possible to hardwire this into the DTD
I'm sure, but probably makes more sense to state it in the tag
library, application guidelines, and perhaps comments within the
DTD itself.

Alvin Pollock
Finding Aid Conversion Specialist
Electronic Text Unit
UC Berkeley Library
[log in to unmask]