Re: Tag Languages, UID's etc.

[duane <duane@xmlglobal.com>]

Curt/all:

Sorry - I should have jumped in sooner on this thread.  Here is a quick
demo.

Here is an XML message.  Most mail clients will not display the
characters properly.

</xml version="1.0" encoding="UTF-16"?>
<ñ�o”Å‚Ì>
  <‘��‡ƒvƒ�ƒfƒ†�[ƒX‰>01</‘��‡ƒvƒ�ƒfƒ†�[ƒX‰>
    <ïŽÐ‚Å‚ ‚è�>01</ïŽÐ‚Å‚ ‚è�>
    <A�Å�V‚ÌCAI>01</A�Å�V‚ÌCAI>
  <‚µ‚Ä�AŠeŽí�»•i‚>12.99</‚µ‚Ä�AŠeŽí�»•i‚>
</ñ�o”Å‚Ì>

A little criptic?  Not only do you have to understand it but your
application might too.  Unfortunately,  your application cannot even
recognize <Element> is really the same as <element> (in enlgish) becuase
to a machine, one letter case insensetive will make the app barf.  Those
of us who occassionally write XML with a text editor spend a lot of time
tracking down small mistakes like this case mismatch too (expecially me
with my Canadian grammar and bad spelling ;-)

Okay - let's try again.  This time with UIDs as fixed attributes of the
DTD:

</xml version="1.0" encoding="UTF-16"?>
<ñ�o”Å‚Ì UID=GXL:10021>
  <‘��‡ƒvƒ�ƒfƒ†�[ƒX‰ UID=GXL:10022>01</‘��‡ƒvƒ�ƒfƒ†�[ƒX‰>
    <ïŽÐ‚Å‚ ‚è� UID=GXL:10023>01</ïŽÐ‚Å‚ ‚è�>
    <A�Å�V‚ÌCAI UID=GXL:10024>01</A�Å�V‚ÌCAI>
  <‚µ‚Ä�AŠeŽí�»•i‚ UID=GXL:10025>12.99</‚µ‚Ä�AŠeŽí�»•i‚>
</ñ�o”Å‚Ì>

But this time a quick Registry lookup (or local cache lookup so we don't
have to keep using the registry) gives us the following xml fragments:

...
<UID>UID:10021</UID>
<Description xml:lang="EN-US">The element container hold a date and
stock quote</Description>
<EquivalentTo YUID="cXML:10023>YourKnownElementName</EquivalentTo>
...
...
<UID>UID:10022</UID>
<Description xml:lang="EN-US">The element container hold the current
month</Description>
<EquivalentTo YUID="cXML:10023>YourKnownElementName</EquivalentTo>
...
UID:10023 is the day element, 10024 the year and 10025 the price of the
stock expressed in Thailand Baht etc.

Now both you and your application can understand the message.  The key
here is that the elements YOU chose to use are always able to convey the
human and sometimes machine knowledge base for the content whiel the UID
will always be primarily a machine reference but can also be used by
humans to grab the true semantics by browsing a registry.

The mechanism is simple and effective and once you and your app
understand the UID,  you don't have to query the registry each time it
comes in or another trading partner requests it.

<SideNote>
I will likely not be able to respond to this thread for 48 hours.
</SideNote>

Cheers!  Great work going on here.

Duane Nickull
Technical Architecture