RE: [Refdb-users] Importing RisX into RefDB

[Couwberghs T. <tco...@sc...>]

Hi all,

I agree with Alan here. He totally makes my point. Also, I'm very
pleased with the RISX-format as it is now and I haven't encountered any
problems, thanks!

Greetings,

Tom Couwberghs

> On Thursday, April 24, 2003, at 11:41  PM, Alan Anderson wrote:
>
> > If you leave risx as simple as possible, i.e. a one-to-one mapping=20
> > between the XML tags and the RIS tags, then define additional DTDs=20
> > as needed for your application--i.e. type of reference. Once the DTD

> > is defined, you create a stylesheet to convert between the two. In=20
> > this way, Bruce can define a "movie" DTD which maps the director to=20
> > A2 and back, and I, someone who's never likely to use a movie=20
> > record, don't need to be confused by 20 or 30 attributes on the=20
> > author tag which I'll never use.
>
> And are you going to create an "edited book" DTD so that you easily=20
> define "editor" and such?  ;-)

I'm not sure I would go to this extreme.


> I don't think this will add much complexity, and it would in fact=20
> maintain a one-to-one mapping of the structure at the element level;=20
> it would simply be that attributes are added in key places like=20
> name/author and maybe title to hint at proper input.
>
> You could have, for an edited book:
>
> <name role=3D"editor" type=3D"personal"> <firstname>Alan</firstname>
> <lastname>Anderson</lastname>
> </name>
>
> or, for a hearing:
>
> <name role=3D"legislative body" type=3D"corporate">United States=20
> Senate</name>
>
> or, for a PCOMM record:
>
> <name role=3D"recipient" type=3D"personal"> =
<firstname>Alan</firstname>
> <lastname>Anderson</lastname>
> </name>

I understand your point, and I agree that this would be beneficial IF we
were creating a new datatype, but the question still remains "who
determines how the name is converted to RIS and stored in the database?"
After all, RefDB is using RIS as its data model.  If all of this logic
is placed in the DTD, you force RefDB to have to convert it.  So, if
RefDB chooses to store the editor "role" into the A2 field in the
database, then we are forced to live with that, and anyone who has been
storing this in A3 will have to convert all of their references.  It
seems to me that if you leave this detail to the user, then RefDB
maintains its simplicity while still providing an elegant and flexible
mechanism for entering records.


> I agree at some point a line needs to be drawn to limit complexity=20
> (and the MODS schema being a more ambitious data model that I'd like=20
> to see future support for), but I don't think there's much point in=20
> XML-ifying RIS without rationalizing it to some degree.

I agree for the need to rationalize.  Given a decision to restructure
RefDB around the new data model defined in the risx DTD, then I think
your suggestion is great.  But to minimize the impact of risx on the
current RefDB, I think you have to consider closely the conversion
between data models. Rationalization for the new data model should
include impact on the current code; I hesitate to suggest drastic
changes to a stable code-base without good reason.  I'm sure Markus is
busy enough.
;)

Al



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf _______________________________________________
Refdb-users mailing list
Ref...@li...
https://lists.sourceforge.net/lists/listinfo/refdb-users