Terminology in XML-Link Draft
Even if we adopt an attribute-remapping scheme, the attributes described
in XML-Link would be studied by many and adopted in DTDs used by
non-hypertext experts. I suggest they be as user-friendly as possible.
I know these issues have already been haggled-over, but there are many
terms and organizational constructs in the XML-Link draft that IMHO could
use a makeover. (Hopefully the ongoing alteration & simplification of
the draft will deem some of these terms unnecessary.)
A few examples of tricky, non-parallel nomenclature that will unnecessarily
o The XML-LINK attribute of a sample <XML-LINK> element with
the fixed value of "LINK."
o "Labels" that go in the TITLE attribute.
o The BEHAVIOR attribute included in the "Behavior" category of
"(3.1) Information Associated With Links."
For section (3.1), I suggest defining three categories for link attributes:
Going into details:
*** SAMPLE ELEMENTS ***
Even with disclaimers such as "note that the element types do not need to
be those given," how about declaring sample element names such as
<MY-XML-LINK> or <MY-SIMPLE-LINK>? This would reinforce to non-HyTime wizards
that you still can name elements what you want.
*** CLASSIFYING ATTRIBUTES ***
The full list of acceptable values for the XML-LINK attribute should be
presented up front in the draft. I suggest not having a value of "LINK." A
more parallel and easy-to-understand set of values:
XML-LINK = ( SIMPLE | EXTENDED | LOCATOR | GROUP | DOCUMENT )
Is the set of pre-defined link roles available somewhere? I seem to have
missed them. These values should definitely be easy for laymen to understand.
Why call it title? LABEL seems comfortable for both links and alt-text for
images. If LABEL is good enough to be defined in "1.3) Terminology"....
*** RESOURCE ATTRIBUTES ***
HRTYPE is a poorly-chosen abbreviation. HREF-TYPE leaves no ambiguity.
(This attribute sounds like its history, though.)
As long as we're using HREF....
implied-locsrc = (referrer | docelem) is the kind of stuff that gives
HyTime a bad name. How about:
DEFAULT-BASE = ( REFERRING-DOC | CURRENT-DOC )
*** BEHAVIORAL ATTRIBUTES ***
The verb "SHOW" seems out of place among its sibling nouns. (INCLUDE |
REPLACE | NEW) seem decent, though one could argue that EMBED is more
precise than INCLUDE (despite it's Netscapial overtones). In past mail,
I see EMBED listed as a value for BEHAVIOR -- is there a list of
probably BEHAVIOR values somewhere?
The term ACTUATE is indeed precise. I just think it's unnecessarily nerdy.
AUTO and USER seem OK -- I prefer AUTOMATIC and MANUAL.
Another alternative for this attribute name: TRIGGER.
I think BEHAVIOR is a great name for the *category*, not for this one
particular attribute. Using the same name for both is confusing.
*** etc. ***
Many of my suggestions have more letters than the terms in the draft.
I lobby that the added legibility is worth it, and given that
many of these values occur in DTDs only, who cares?
Speaking of DTDs, in Microsoft's CDF document they refer to the chosen element
names as making up a "profile." This term is a step up in accessibility from
DTD. I vote that a universal synonym for DTD/Document Type Definition
suitable for Average Folk to understand will help market XML. Some
suggestions as an alternate to Profile:
Tag Set (Yes, elements are not tags. If it sells, I don't care.)
XML Tag Collection (XTC, for short ;-)
XML Tag Definitions (XTD -- nice 'n' close to DTD)
Docunition (Short for Document Definition -- just kiddin' ;-)
I'd also like to add that "XML" has proved to be a well-chosen name,
suitable for mass-marketing. Has Microsoft registered "ActiveXML" yet?
The Wall Street Journal Interactive Edition
firstname.lastname@example.org phone: 609 520 7361
http://wsj.com fax: 609 520 7137