Re: Undocumented, undiscussed substantive changes in the 10 May draft

I’m also very against the description of @-marked data as “unstructured”. The whole point of attributes is to be a structural unit, representing information in a way that complements the main structural unit (the element). If “unstructured” is meant to mean “non-hierarchical”, because attributes are attached to a hierarchical unit but cannot themselves contain other units in a hierarchical configuration, then maybe I can see the value of expressing something like that. But not with the word “unstructured”.

In addition to that, so far most of us seem to have assumed that iXML is intended to function by using XML as an interchange format (i.e. producing a “visible XML” representation of the input's implicit structure, which can used for whatever purpose one wishes). That seems to me to be what the spec has said for as long as I’ve been involved, and it’s certainly the basis upon which I’ve been building my implementation. If that’s *not* what’s going on, in the consensus view of the group, I think we need:

a) a detailed discussion of how the misunderstanding has come about, so that we can all be clearer about what exactly this technology is meant to be, and so that anyone who’s implemented (or begun implementing) iXML can ensure that their implementation is future-proof;
b) a discussion of what’s in the spec - my interpretation of the spec (the version we’ve known up until now, in any case) is that XML is necessarily at the core of iXML. If that isn’t the case - for example, if future versions of the spec could do away with XML entirely without its being seen as a fundamental change in the nature of iXML - then I would like a discussion of that. I don’t feel that such a reading is possible in the spec with which I’m familiar, but if I’m misreading or misunderstanding, I’d like to know that, and to suggest revisions to ensure nobody else makes my mistake.

All very best,

BTW

___________________________________________________ 
Dr. Bethan Tovey-Walsh 
Myfyrwraig PhD | PhD Student CorCenCC 
Prifysgol Abertawe | Swansea University 
Croeso i chi ysgrifennu ataf yn y Gymraeg.

> On 10 May 2022, at 17:43, Norm Tovey-Walsh <norm@saxonica.com> wrote:
> 
> Hello,
> 
> The 10 May 2022 version of the Invisible XML specification has been
> redrafted with new language that is, I believe, a significant step
> backwards in terms of readability and understandability.
> 
> For example, in the 7 April draft, under rules, we find:
> 
>  A mark is one of @, ^ or -, and indicates whether the item so marked
>  will be serialised as an attribute (@), an element with its children
>  (^), which is the default, or only its children (-).
> 
> In the 10 May draft, it reads:
> 
>  A mark is one of ^, @ or -, and indicates whether the item so marked
>  will be serialised as a structured element with its children (^) which
>  is the default, as unstructured data in an attribute (@), or deleted,
>  so that only its children are serialized (-).
> 
> That’s just one example of a pervasive move away from concrete
> descriptions of the XML serialization to favor wishy-washy (and
> undefined) concepts such as “structured” and “unstructured”. Those words
> do not rise to the level of terms of art that can be used in a technical
> specification without definition.
> 
> I object to this change having been silently made with no corresponding
> issue or discussion. This is not merely an editorial change, this
> represents a departure in technical clarity and perhaps even underlying
> meaning from the previous drafts.
> 
>                                        Be seeing you,
>                                          norm
> 
> --
> Norm Tovey-Walsh
> Saxonica