> What are these different ways of defining nodes called? And when is it 
> good to use the one instead of the other?

An element which does not have content (whether or not it has 
attributes) may be expressed either with an open and close tag pair:
	<foo></foo>
or with an "empty element" tag
	<foo/>

These two forms are completely identical semantically; the difference is 
almost purely one of convenience and personal preference.

Why "almost"? Well, it is officially preferred (ie, there is a SHOULD 
keyword in the XML Recommendation) that the empty element tag be used 
only for elements whose DTD or Schema specification says they can never 
contain content (ie, that in a Valid document they will always be 
empty). But I don't believe anyone has ever actually followed that 
preference, since the empty-element tag is a trifle more efficient to 
type by hand or process by machine. So that proposed convention seems to 
have died due to lack of support from the community, and in practice the 
two forms really are completely equivalent.

I'm presuming that's what you meant by "open" versus "closed".

Note that it's completely orthogonal to the decision of whether the 
element should carry data as attributes, as content, or both. That's a 
decision made when the language which XML is being used to implement is 
designed, based on a combination of stylistic preferences and practical 
concerns. Attributes have their values normalized to some degree, have 
no meaningful order, and are limited to name-value pairs, whereas 
contained nodes may include any desired mix of 
text/elements/comments/processing instructions, do not normalize the 
text, and do reliably report their order as a possibly significant piece 
of information.  Both approaches have advantages and disadvantages. 
Generally, by convention, attributes are used for information that 
modifies how the element they're attached to should be interpreted, 
while the information that is logically contained by that element is 
placed in its content ("child nodes")... but there are exceptions, which 
is where the stylistic element comes in. Do what makes sense given what 
the document is intended to mean, but if there's any doubt you should 
probably lean toward child nodes -- they give you more options for 
future enhancement. If you need more guidance on this, websearching for 
"XML children versus attributes" or similar phrases will find many other 
discussions of this trade-off.

Hope that helps.