[sc34wg3] XLink support in XTM

[Murray Altheim]

Quoting Lars Marius Garshol <larsga at ontopia.net>:
>
> * Murray Altheim
>>
>> We gain linking. Absent XLink we have no expression of what happens
>> with links: how they're defined, the model behind them, their
>> behaviours, etc.
>
> What do we need this for? We reference the URI specification, which
> gives us the definition of URIs, the model behind URIs, and the
> behaviours of URIs. What does XLink add to this that's of any use to
> us? Where would it be used, and for what? Until you can even attempt
> to explain this, why should I listen to you? (And don't tell me
> you've attempted to explain this, because you haven't.)

If you have read through the URI specification you would accept
that it does not provide a definition of linking, about linking
behaviour, and certainly nothing about providing a link syntax
that connects to that behaviour. The IETF RFC states:

    Uniform Resource Identifiers (URI) provide a simple and
    extensible means for identifying a resource.

URIs are only a identifier not a link. Like the current XTM 2.0
draft, the URI RFC doesn't even include the word "link". URIs are
not links even in use (i.e., they're called "URI References").

Linking is the action. I shouldn't have to explain this. That you
point to URIs as your solution to this either shows your ignorance
(which I doubt) or your unwillingness to accept the need for a
linking model. If you think I'm being obstinate, imagine my feelings
about how you keep harping on me to explain what I mean by a model.
XLink *is* a linking model. Forget the word "model" if you like.
What XLink provides for any XML syntax is pretty freakin' obvious
to anyone, and is very clearly stated in the W3C XLink Recommenda-
tion. Its very existence as a specification states its importance:
it provides both an XML link syntax and a complete picture of how
linking behaviours are expected in applications that use that syntax.
It's hard for me to be any clearer than this, and I've been saying
the same thing for a very long time. These were the very same
arguments that were used to include XLink the XTM 1.0 specification
six years ago. This is not new information. I can keep repeating it
if you like, but you obviously don't like.

As to listening to me, if we've decide to begin characterizing each
other's communications, I'd have to say that you have clearly from
the beginning have shown a decided lack of interest in listening to
anyone. Sure, you reply, en masse, but you seldom seem to respond,
just rehash old arguments without actually replying. Just as with
the <mergeMap> discussion, on at least several occasions, myself and
several others have provided replies; you simply don't like them
and generally ignore their them (or the substance of the argument),
choosing instead to throw out a comment, wait awhile, then state
that there isn't "consensus" on the issue and later restart the
conversation, and then demand that we reengage with you. At some
point people simply grow weary of replying. If I no longer put the
energy you demand into these discussions it's because I long ago
learned that your strategy has always been to wear people down with
persistent and consistent emails, and then when people grow tired
you have the nerve to accuse them of not providing you with what you
want. Not everyone has the time to waste on such nonsense.

This community has suffered enormously because of your abusiveness.
You probably don't see it that way, but you might note there's not a
lot of people bothering to comment on this anymore, not many of the
people who were there in 2000. Why? Figure it out. It's not because
of me. I don't hang around here enjoying the show, and contrary to
what you might think I take no pleasure in annoying you. If I don't
leave it's only because I happen to both care about the Topic Maps
activity and what remains of its community, and also that leaving
would be a victory for you, that you've finally wore me out. But
don't expect me to put in the energy of a full-on conflict. If you've
turned me into a harrier, that's not my fault. other than that, read
Sun Tzu for an explanation.

> In this conversation there has been a clear pattern from the very
> beginning. It's all along been obvious that you are complaining about
> the contents of documents you have done no more than merely skim
> through.

I have read each of the drafts completely.

If there's any clear pattern it's that you're willing to be abusive
once people don't do what you want, don't fulfill your demands to
continue in an pointless battle.

> In addition to this, your complaints have been free of any substance.

We'll let others judge that, since you seem perfectly ill-equipped
to do so.

And to note, my complaints were committed with more care and
substance a year or so ago when I cared more about the outcome or
thought there might be some chance that the energy I expended had
any impact. But I also learned long ago that you are unwilling to
alter your opinion on anything based on outside influence, unless
you already happened to agree, or perhaps if someone is willing to
spend countless hours arguing with you, but I've seen little
evidence of that. This activity has long spun out of control, and
I don't have the energy, the time, or the ability to fix it.

So it's not a lack of substance, it's a lack of commitment to argue
with you anymore, for despair of there being a point. What's that
quote about insanity being expecting a change of output with no
change of input?

> You complain that we lack a "linking model", and yet you  have
> persistently refused to explain what it is.

I've explained it before, and I've persistently pointed at the
existing one for XML, i.e., XLink.

I don't know why I should be required to explain what XLink is or
why it's valuable. That should seem patently obvious to anyone
who knows anything about XML. If it's not being taken up within
the W3C's RDF community it's because XLink and RDF are incompatible,
but that can't be said for XTM, as it's been in use for five years,
successfully. I note that there is a SourceForge XLink processor
(XBRLAPI), one of the Topic Map vendors (Empolis) has an XLink
engine (X2X), Fujitsu has a full XLink/XPointer implementation,
and the Mozilla browser supports simple XLinks (the kind used in
XTM 1.0). If there is any question of the value of XLink as a
specification, read some of those project/product descriptions.

But even if you decide not to use XLink for XTM 2.0 you will still
need a description of what I've repeatedly stated, i.e., where
there are inherent links in the model, how they map into the
interchange syntax, the expected behaviour of those links under
the various contexts of their uses. If you don't want to call
that a model fine. But you haven't even bothered to describe this
for XTM, an application that is *full* of links. It's an omission
that doesn't bear scrutiny. That you have the nerve to get on my
case about it is only surprising in its continuance.

> Instead, I am treated to sarcastic remarks about how the last XTM 2.0
> draft, the result of five years of work, doesn't do linking, "ha ha
> ha". I can only conclude that you are resorting to sarcasm because
> you have nothing else to offer.

Sarcasm? I was pointing out that in the entire XTM 2.0 draft the
work "link" or "linking" doesn't exist except to point out that it
doesn't exist. And yes, I have nothing *new* to offer -- I keep
repeating the same thing, you've just not countered with any reply
that explains why XTM doesn't need to define its linking if it
stops using XLink.

> You should also note that whether we use XLink or not in XTM, the
> only thing from a <resourceRef/> element that survives into the TMDM
> is the URI.

I'm not commenting on the TMDM, I'm commenting on XTM 2.0. I could
give a squat about what's in the TMDM. But if neither the model for
Topic Maps* nor the XML syntax expression for use of Topic Maps on
the Web, XTM, nor any specification associated with the entirety of
the definition of Topic Maps contains a definition of what links
are, where they are and why, how they're expressed in syntax and how
applications are expected to handle them, well, well, that would
seem a giant omission. But apparently you don't think so, so I can
generally expect that the status quo will remain, no matter how
many remarks I make, snippy, substantive or not.

* which I wouldn't expect necessarily itself contains a definition
  of linking, unless it's wrapped into the model itself rather than
  a separate linking model, plainly not the case here.

> The upshot is that even if using the XLink attributes  instead of our
> own attributes added something, that something would  be lost once
> the TMDM instance has been built. So any definition,  model, or
> behaviour that XLink would offer us would not make any  difference on
> the topic map represented by the XTM document. This has been the same
> since the first TMDM draft in June 2001, and nobody  (you included)
> has ever complained.

Again, I'm not discussing the TMDM, I'm discussing XTM. I would see
no point in complaining about a lack of linking in the TMDM, as I
wouldn't have expected to see linking (necessarily) there. But there's
an omission I have kept on about, and that's in the definition of
Topic Maps as it translates into a version for the web, in XML, i.e.,
XTM, you had better have a definition of linking or you don't have
linking. Pointing at the URI spec is weak, weak, weak. You don't get
linking with URIs, you get references. References don't *DO* anything.

>> This continued business about having to declare a namespace (which
>> happens once per document and amounts to 42 characters) I find
>> rather silly, a complaint over nothing.
>
> The complaint is more than that, and you know it.

Then why keep bringing it up as a namespace issue as if that's
the biggest complaint? Unless the other arguments in this case
seem to lack any substance I can't see a reason.

>> Yes, and I think you need to reread my postings on this. I provided
>> responses to this, and have repeatedly posted justifications for the
>> initial choice of XLink as well as its continued use.
>
> Where? I haven't seen anything with any substance to it.

I haven't seen anything of substance in your emails either, just the
relentless demand for counterargument, such as requiring people who
disagree with you to prove their argument, not you. Your opinion of
your own remarks' substance isn't justified by them. They're just
statements of opinion like anyone else's. You just make them over
and over until you believe them and expect everyone else to as well,
or at least accede to them unless they're willing to engage you in
endless battle.

> * Lars Marius Garshol
>>
>> I've asked you before what XTM needs a linking model for, but have
>> so far received no reply.
>
> * Murray Altheim
>>
>> As I mentioned above, I think you need to reread my postings on this,
>> as well as those of others.
>
> So far I've seen one sentence from you on this, and that sentence is
> quoted above. Conal Tuohy did post on the subject (and you replied to
>  him in a way that seemed to suggest that what he wrote was what
> you'd  meant), and I've replied to Conal's posting. That's the sum
> total of  what I've seen on the subject of what a "linking model" is,
> and how  it might apply to XTM.

Yes, Conal and I meant basically the same thing. And this
discussion goes back *years*, back to 2000. If that's the
sum total of what you've seen on linking, you've got either
a short or very selective memory.

[...]
> XTM 1.0 is a specfication which such profound problems that it's hard
> to see how it could be used as an example to justify anything. We've
> worked for the past five years to solve the problems created by that
> specification, and if you don't like the result, the least you can
> do is to tell us why.

XTM 1.0 is a specification that some people are still proud of, despite
your opinion on the matter. It was created by a group of the some of
the most intelligent people I've ever met, doing in one year what you
haven't been able to do in five. You think you're such an intelligent
engineer, but we'll see how history treats XTM 2.0, or if you haven't
created the ugly, unusable monster that it's looking to be.

As to telling you why I don't like XTM 2.0, what the hell do you think
I've been doing for the past few years? Oh, don't answer: writing
snippy little substance-less emails for my own jolly entertainment.

[...]
>> If you are comfortable with an XTM that doesn't include any linking,
>> don't expect XTM 2.0 instances to do any linking, nor see any reason
>> why XTM 2.0 needs linking, then hell, leave any definition of linking
>> out. Yeah, actually, if you don't see the need to define linking I
>> can completely agree with you -- don't define any linking.
>
> That's what we've done, and that's what you seem not to like. Maybe
> you could explain why you think this is a problem? And if you can't
> explain why you consider it a problem, maybe you can shut up on the
> subject? Having you repeat over and over again for months that we
> don't have a "linking model" without ever bothering to explain what
> it is or why that would be a problem is beginning to grate on me in a
> serious way.

I have explained in very simple fashion, repeatedly, that the
specification for an application that expects to use linking, that
is, expressions of links between documents and document components
that are expected to be acted upon by software applications must
have an expression of a link syntax, how that syntax is to be
interpreted by software, yadda yadda, precisely what XLink is and
does.

As to grating on you, I really could care less. What are you going
to do, be *more* abusive?

> In case it was not clear from the above: you are not being asked
> to shut up. You are asked to either state
>
>   a) what a linking model is (more than one sentence, please),
>
>   b) why XTM 2.0 needs one
>
> *or* shut up.

I think this pretty much demonstrates the problem you have been
to the health of this community.  When if ever I decide to shut
up be well assured it won't be because of any demand of yours.

Murray

...........................................................................
Murray Altheim <murray06 at altheim.com>                              ===  = =
http://www.altheim.com/murray/                                     = =  ===
SGML Grease Monkey, Banjo Player, Wantanabe Zen Monk               = =  = =

      In the evening
      The rice leaves in the garden
      Rustle in the autumn wind
      That blows through my reed hut.  -- Minamoto no Tsunenobu