[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: a report on any xml file, what information is useful?

• From: ALT Mobile <dev@a...>
• To: Len Bullard <cbullard@h...>
• Date: Mon, 11 Sep 2006 08:08:11 -0400

Those simple screen shots were related to XML analytic functions: 
specifically interactive namespace reporting and document statistics.

The "visualization fosters comprehension" theme is evident throughout 
the XML development life cycle. It is less apparent during the editing 
and viewing of small documents where expressing XML in its serialized 
form (that is, text) is more appropriate. More specifically, only when 
you forgo whitespace issues such as indentation and canonicalization and 
forgo printing needs is this approach to viewing XML adequate.

For the vast majority, let's say 90% of XML use, the most appropriate 
means to understand an XML document/system is to employ visualization 
technologies.

Consider the following use cases:

1. Server generated XML which usually contains mid-to-large XML data 
sets. Whether the schema is or is not under your control or even 
available, downstream developers consuming the XML will have understand 
the data. Based on our research, the most successful approach for 
application developers is to first understand the document structure and 
specifically the container-contained (parent-child) node relationship. 
For architects, the first requirement is document and element sizing 
issues as they drive technology decisions such as parsing, searching, 
and mapping concerns. So an application developer needs a "lay of the 
land" macro view of the document such as seen here:

http://altmobile.com/sample_shots/1157968056390.png

and through a subsequent navigation:

http://altmobile.com/sample_shots/1157968285093.png

The traditional text centric view or simple tree view is usually 
inadequate as you have to constantly scroll in both the horizontal and 
vertical directions and view non-essential data such as attribute values.

In the macro view, it is not important to see schemas and attributes but 
rather element relationships without having to constantly scroll. User 
interface researchers are in broad agreement that hyperbolic views are 
best in these cases as they show the "context and focus" and allow your 
eye to easily see the transition from one node cluster to another 
without the typical "jump" that occurs when clicking on the scroll bar.

Furthermore in traditional text/tree views the XML data is presented in 
a vertical manner and the eye is unable to transmit all of the data 
needed to form a mental model of the data. Empirically you can tell when 
this happens to developer when you see his eyes glaze over.

For us, we married hyperbolic tree technology with the DOM to create the 
HyperDOM browser enabling node navigation rather than window scrolling.


2. Monitoring web service data on the wire. In this case, it is 
important to correlate both the request and response. A classical text 
centric view can be seen here:

http://altmobile.com/sample_shots/http-monitor_std_view_1.png

where we cannot even see the entire document without scrolling and are 
easily lost. The XML visualization approach is shown here:

http://altmobile.com/sample_shots/http-monitor_browse_view_1.png

and here:

http://altmobile.com/sample_shots/1157964672265.png

where we can visualize several HTTP transactions.

3. Viewing SOAP with Attachments (SWA) data. Whether your tool 
decomposes the SWA mime document like this:

http://altmobile.com/sample_shots/swa_editor_1.png

or this:

http://altmobile.com/sample_shots/swa_hyperdom_1.png

clearly visualization is needed to understand the data. A text editor is 
the wrong way to go and accordingly traditional XML tools will not 
support SWA, it's variants, binary XML or many of the security XML 
technologies.

4. Printing XML documents. We call this "paper-based visualization" and 
it is amazing to see that QA and integration engineers need to add 
written annotations to XML documents. Text print outs and vertical trees 
do not provide enough space to write comments as seen here:

http://altmobile.com/sample_shots/1157974426125.png


But horizontal trees do, as seen here:

http://altmobile.com/sample_shots/1157974326359.png


If you look at the life cycle of XML data-- not just editing-- you will 
see that use of visualization technologies dramatically increase 
knowledge of the XML data model and the specific document instances. 
Programming errors are reduced because you can inspect your variables 
and get instant feedback. For us that's using our embedded Java language 
API with your IDE or our Web Services API for non-Java languages.

Once you can see your data, you can understand your data.

After using visualization and object-oriented user interfaces to build 
XML systems, you start to realize that XML visualization tools compare 
to text editors just like graphical user interfaces compare to command 
line prompts. They both have a place but GUIs and visualization tools 
are usually more appropriate.


For more information on our XML Visualisation technologies:

* DOM Browser and HyperDOM:
http://altmobile.com/Technologies/Enterprise%20XML/Visualization/Visualization.html

* Web Services data visualization:
http://altmobile.com/Technologies/Enterprise%20XML/HTTP%20Monitor/HTTP%20Monitor.html

* SOAP with Attachments (SWA) Visualization:
http://altmobile.com/Technologies/Enterprise%20XML/SOAP%20and%20SOAP%20With%20Attachments%20Processing/SOAP%20and%20SOAP%20With%20Attachments%20Processing.html


I hope this make it a bit more clear.


--Zaid

http://altmobile.com



Len Bullard wrote:
> Interesting discussion.
>
> What is the basis for the claim that the reports in your jpgs represent a
> 'visualization' that fosters comprehension?  Or is there a different
> visualization for which you didn't provide an example?
>
> len
>
>
> From: ALT Mobile [mailto:dev@a...] 
>
>
> Higher-level XML features such as analytics and differencing are harder 
> to implement in editing tools because of the text-centric nature of the 
> user interface. This is because the user has to constantly change 
> focus/views. For this reason, we implemented our XML tools on a 
> visualization core and are able to leverage all of the innovations in 
> direct manipulation and object oriented user interface technologies.
>
> As we like to say in describing the importance of XML visualisation and 
> XML analytics:
>
> "Visualization fosters comprehension and analytics fosters insight"
>
>
>   

• References:
  • RE: a report on any xml file, what information is useful?
    • From: "Len Bullard" <cbullard@h...>