[tei-council] Disambiguation of <ident> and <idno> (and also <gi>)

Martin Holmes mholmes at uvic.ca
Fri Nov 25 16:06:41 EST 2011

We seem to be a long way from a consensus here, in that most people 
haven't expressed a view, and those of us that have seem to fall into 
two main camps: those who believe that <ident> and <idno> overlap 
substantially, and those who would like them not to, and want to word 
the guidelines accordingly.

However, nobody has actually objected to the wording that I proposed 
adding to the existing element definitions:

<ident> should be used for tokens such as variable names, class names,
type names, function names etc. in formal programming languages. (It
should not be used for element and attribute names in XML, for which the 
special elements <gi> and <att> are provided.)

<idno> should be used for labels which uniquely identify an object or
concept in a formal cataloguing system such as a database or an RDF
store, or in a distributed system such as the World Wide Web.

So unless someone does voice a spirited objection in the next little 
while, I propose adding those to the two <elementSpec>s. That pushes us 
oh-so-gently in the direction of making more of a distinction between 
the use-cases for the two elements. I'm then going to look at the text 
of these two sections:



to see if further disambiguation is possible or desirable in the text.

Incidentally, while I'm here: anyone know why the phrase "22 
Documentation Elements" is not a link to the chapter, in the <ident> ref 



On 11-11-24 06:41 AM, Gabriel Bodard wrote:
> I think I support Martin's position here, that the distinction between
> <idno>  and<ident>  is as he described it in his proposed revisions to
> the guidelines and is pretty clear and unambiguous. The difference is
> also one of context, in that<idno>  provides an identifier to its parent
> element, while<ident>  identifies a string in prose or elsewhere as some
> kind of name or identifier. The two are usually very different kinds of
> animal (as he describes below).
> Might they overlap sometimes? Maybe. Well, that's probably inevitable
> even, but I'm not sure I'm entirely comfortable with the idea of using
> <ident>  to tag an isbn used outside of the context of a bibliography.
> (Even if we do decide that works, though, I don't think that destroys
> Martin's definitions; it would just be an edge case where we see
> overlap, but where context makes it clear what's needed.)
> G
> On 2011-11-24 13:43, Martin Holmes wrote:
>> If everyone is of the view that<idno>   and<ident>   overlap, and the same
>> information belongs in one or the other depending on context, then I'll
>> abandon my attempt to clarify their definitions; it was based on a
>> misunderstanding. I'm with Laurent in believing that they are for
>> completely different purposes, but if Council believes they're not, then
>> we should deprecate one of them because the difference is pointless and
>> confusing. I thought I'd identified a useful distinction that justified
>> the existence of both, but if not, then I'll just close that ticket and
>> raise another suggesting that one be deprecated.
>> So can we have a show of hands: Who agrees with the following statement?
>> The same information can appear in both<idno>   and<ident>, and the
>> decision as to which to use depends entirely upon context.
>> -1 from me.
>> Cheers,
>> Martin
>> On 11-11-24 04:51 AM, Lou Burnard wrote:
>>> On 24/11/11 12:42, Laurent Romary wrote:
>>>> Le 24 nov. 2011 à 13:39, Sebastian Rahtz a écrit :
>>>>> On 24 Nov 2011, at 11:48, Laurent Romary wrote:
>>>>>>> just what<ident>     is for, surely? it is sugar for<hi rend="identifier">
>>>>>> That's exactly what I (didn't - indeed) want (ed) to hear from you. From an ontological point of view I don't like so much to see the same object receiving a schizophrenic treatment depending on where it appears (and we do mean the same thing in both cases).
>>>>> <idno>     is a superset of<ident>.  It does the same job, but adds extra semantics, viz that
>>>>> not only am I identifying this as an ID in general, but actually as the actual ID of my parent object.
>>>> Should they then have the same content model (ident becoming recursive if Idno is)?
>>> Hard to say no to that... tho I'd still rather see<hi>    permitted in
>>> there....

Martin Holmes
University of Victoria Humanities Computing and Media Centre
(mholmes at uvic.ca)

More information about the tei-council mailing list