Bill,
   A folder is a physical unit.  A file is a structural component.  Do you
actually mean that you have a component that is a file?   If so, one
would select the value "file" for the "level" attribute for <c>.

   <unitloc> is for tagging physical locations such a box numbers or
repository locations.  That's why it's not available within <unittitle>,
a structural element.

Michael Fox

 On Thu, 6 Mar 1997 22:16:09 -0500,
Bill Landis  <[log in to unmask]> wrote:

>Does anyone have any thoughts on what the difference between the
>following might be?  I'm referring to how one might best represent
>information about a component which is a folder.
>
>Example 1-
><c0x level=unspecified>
>   <did>
>      <unitloc loctype=container containert=folder></unitloc>
>      <unittitle>Academic Affairs, General</unittitle>
>      <unitdate type=inclusive>1984-1990</unitdate>
>   </did>
></c0x>
>
>Example 2-
><c0x level=unspecified>
>   <did>
>      <unitloc loctype=container containert=folder>
>           <title>Academic Affairs, General
>                <date type=inclusive>1984-1990</date>
>           </title>
>      </unitloc>
>   </did>
></c0x>
>
>These seem to me to be the 2 ways that the DTD allows one to handle
>calling a component a "folder" and giving it a name and a date.  Neither
>seems like a good solution to me and I'm wondering if someone might be
>able to enlighten me a bit.
>
>In Example 1, the tagger is forced to use <unitloc> to establish that
>this component is a "folder", but essentially has to leave the tag empty,
>since <unitloc> can't contain <unittitle> or <unitdate>.  Is that
>correct?  Is it "legal" to use a tag that surrounds no text?  If SGML is
>supposed to make text "smart" this approach to getting the information
>that a component is a folder into the <did> seems wrongheaded to me.
>
>In Example 2 the tagger can embed the label and extent date of a folder
>inside of the tag identifying that component as such.  This seems (in my
>admittedly limited knowledge) to be a truer representation of what SGML
>ought to be able to do (take advantage of nested information) than
>Example 1, where the relationships between the tag establishing the fact
>that a component is a folder, the label of the folder, and the extent
>date for the materials contained in the folder are established solely by
>these tags all being contained within the same <did>.  But I'm not sure
>if, in Example 2, the use of the generic <title> and <date> don't loose
>me some of the power of being able to identify label/date information
>about a component unit.
>
>I'm wondering why it is that <unittitle>/<unitdate> can't be contained
>within <unitloc> if it is this tag that established the fact that a
>particular unit is a folder, etc.?  And if my 2nd example isn't a good
>approach, does anyone have any thoughts on when being able to embed
><title> and <date> inside of <unitloc> might be useful?  Is <unitloc>
>only for tagging folder/box/etc. numbering information and, if so, what
>should a repository do when they don't number folders?
>
>This may have been addressed in the past on this list, but I'm pondering
>over how to proceed and can't find anything relevant in my saved EAD list
>messages.  Any thoughts on this would be appreciated!
>
>Bill
>
>-_-_-_-_-_-_-_-Bill Landis | JSTOR Production Coordinator-_-_-_-_-_-_-_-_-_
>
>"All landscapes eventually turn to land,
>        the gold of the imagination to the lead of the reality."
>V.S. Naipaul | The Mimic Men
>                                [log in to unmask] | 313 936.2363-_-
Effective March 15, members of the Processing Department can
be reached at individual e-mail accounts.  The naming convention is
[log in to unmask]      e.g.

[log in to unmask]
[log in to unmask]
[log in to unmask]

Processing Department
Minnesota Historical Society
[log in to unmask] or [log in to unmask]
fax  612.296.9961