Edit and display XML in an efficient way

Question

Hello,
for quite some time now I searched the internet, this forum and some other forums but I couldn't really find an answer.
My first question is, which 'technology' should one use to build an XML-Editor, i.e. how can XML-files be efficiently loaded, displayed, in a JTree or JTable, and edited (i.e. add and remove elements and save the XML-file again)?
The only 'technology' I can think of is DOM/JDOM (that's what I've read on many websites) because one needs the XML-file in memory to edit it. So SAX doesn't help here, does it? But DOM/JDOM is very inefficient because the DOM-tree has to be build and is quite big in memory, at least it's said to be, because I have XML-files which are several MB big. And actually I want to display more than one file in the application (e.g. with a JTabbedPane) so that they can be displayed and edited simultanously. But that means the tree for each file has to be in memory so, I thought this would consume to much memory. And I think it's not a good idea to write the file back when it's not displayed.
Another thing is that I want to be able to search the XML-file so, I think, it's only possible with DOM/JDOM because you can use XPath. I don't know how to search an XML-file fast with an other 'technology'.
If the XML-Schema is know, I could use JAXB, but then I don't really know how to search the XML.
Which 'technology' do the 'professional' XML-editors use?
My 2. question is, if there's maybe an library etc. that already implements editing and displaying for XML-files, which one can use in an application.
I'd really appreciate any ideas.

Answer 1

What you could do is use a SAX parser to parse the XML document and build the JTree. Tree the JTree like an XMl Tree. when the user want yo save the Tree, then write the Tree out to an XML file. THere is no need to have the DOm for the XML; however, to let user have the ability to perrform XPATH, you have to write code for it.
14 MB is not a lot, so i would stick with using a JTree and have each TreeNode associated with an Element in the DOM. Parser are pretty fast..it shoiuld probably not takes more than one or two seconds to parse a 14 MB XMl file. Now, if the user have a 100+ MB file, then it may become a problem. In all, your application will probably require at least 100MB just to render the XML tree.
You can also write a class that will parse the XML document...and serialize the DOM to a file. you can break up the dOM into sevral document object (and store them in a file)..to allow you to traverse faster. Your application (class) will read in the necessary document and work with it. THere are two problem here:
1. IO operations are very expensive. however, most of the time, the application will not be calling for the document file. It takes time for user to type than it would for the application to save user update to a file..so during this time, a Thread can be spawn to perform updating the document (synchronization required)
2. The code is complex..you have to make sure the serialized document is updated. you have to come up with an algorithm (design) on fetching these document object, etc..
Many XML editor I see use the simplier approach..loading the whole XML document to a DOM. .. so 100MB document will require 100+ MB of memory space available..This approach is easy..fast..bt require memory.
There is a trade off..speed, easy vs memory.
It's upto you to determine which path you want to go..you can't have everything.