I was working with "xmlgen-1.4" and had quite a bit of success with it.

However, formatting of the results seemed to be a real pain, and the documentation, examples, disucssion thread, direct discussions with the originator of xmlgen did not seem to help me grasp the issues.  From what I could tell, others in the community were having similar issues.

So to get my job done, I wrote the following routine. (don't laugh - it could be better, but time has not been on my side).

The following code does a very nice job of structuring resulting XML.  I hope you find it as useful as I have.

------
The proc "CleanUP" needs the string of XML that will be cleaned up as the submission variable.  This application generates a textbox to display the results.


    ### XML CLEANUP ROUTINES
    
    proc placeData {indentlevel item} {
		set tabs ""
		set i 1
		while {$i <= $indentlevel} {
			lappend tabs "\t"
			incr i
		}
		if {[llength $item] > 0 } {
			# Write data to the non-visible textbox
			.tb2 insert end "[[join $tabs]]$item\n"
		}
    } ; # end placeData
    
    
    proc CleanUp {xml} {
	catch {destroy .tb2}
	text .tb2
	
	# Gather XML Content
	set indentlevel 0
	foreach item [[split $xml \n]] {
		set item [[string trim $item]]
		switch -glob -- $item {
			<?xml* {
				# Header -- NO indent Adjustment
				placeData $indentlevel $item
			}
			*/> {
				# Handle NULL Value Tags
			}
			</* {
				# End Tag
				incr indentlevel -1
				placeData $indentlevel $item
			}
			<* {
				# Look for tags on the same line verses start of a XML nest structure
				# Get the tag name
				regsub -all {<|>} $item { } tempitem
				set tag [lindex $tempitem 0]
				if [[string match *</$tag> $item]] {
					# In-Line Tag - NO indent Adjustment
					placeData $indentlevel $item
				} else {
					# Tag is head of a super structure
					placeData $indentlevel $item
					incr indentlevel
				}
			}
		} ; # end switch
	} ; # end foreach
	return [[.tb2 get 0.0 end]]
    } ; # end CleanUp