"Prof J D Patrick" writes
>Malcolm, I think that the TIPSTER approach gives a flexibility not
>available in SGML; however I don't think that is the whole of the issue
I think this is like comparing apples with fresh-mown hay. The TIPSTER
architecture is a particular approach to solving a general purpose
encoding problem -- how do you mark certain textual features within a
document when they are (a) transient (b) not necessarily constituent
features of the document. You could use an SGML approach to solve the
same problem and you'd end up with something like what the TEI does with
extended pointers, and what HyTime also does. Either way you'd be using
>that could warrant discussion. SGML (or at least HTML its predecessor)
This is historically inaccurate. SGML was being defined while HTML was a
glint in TBL's eye. ISO 8879 is 10 years old this year. HTML is not yet,
so far as I know, standardized at all. More significantly, HTML is (now
at least) an application of SGML so can hardly have preceded it.
>appeared because of a desire to have a "communication standard" rather
>than a "storage standard". That is, the original intention has been
>usurped for other purposes and hence it (SGML) doesn't do what many
>people might want if they sat down and designed a "storage standard".
What evidence do you have for this extraordinary assertion? As far as I
am aware, SGML allows one to defines both a storage model and a logical
or structural model of a document.
>I for one don't see the need for a "storage standard" which is in effect
>what TEI is aiming for (perhaps subliminally),
>that is, TEI is bringing about the conversion of a "message passing"
>protocol to a "storage standard" for particular domains,
>knocking them off one at a time
>so to speak.
Eh? I don't know how one subliminally aims for anything. As far as I
know, the TEI application of SGML, like the HTML one, is intended to
make it easier for people to communicate. It defines a "message passing
protocol" in the sense that it defines an interchange format. It doesn't
require you to use any particular storage format: though obviously if
you use the same format for storage as you do for interchange there will
be less work to do when data comes in or out of your store.
>I favour a slightly variant scenario that will try to get the
>best of both worlds and we here are working on an implementation for our
>domain, ie. dictionaries. We have stored our dictionaries in structures that
>best serve our purposes and consider them hidden from the view of the ordinary
>user. However if a user wants to talk to our dbases in a nonDBMS way (that is
>not through an SQL interface) then they can do it via a Generalised Dictionary
>Query Language, in our case implemented as a WEB form or template. The
This description sounds to me pretty much exactly what SGML in general
and the TEI in particular is intended to facilitate!