Links to the paper and related texts are at the page DevReports.
These comments appeared during informal discussions of the paper. Comments are reproduced randomly and in a very short form, probably only for internal use. Most comments come from 
Martin Bravenboer. Thank you, Martin!
"What's new?" should be stated very explicitely.
Papers are written not for readers, but for reviewers. The goal of a paper is to be published. He-he.
The paper would better presented at a workshop, not at a conference.
General. XPath example, more examples of DSL implementations on top of Scheme. 2-3 examples.
Near the end of the section "5. SXML, the Lisp Approach". Too much existing work.
"6. XML Virtual Machine", the paragraph: "The way of implementing a given XML standard..." The question mark in a cirle.
"6. XML Virtual Machine", the paragraph: "For our developments we decided to use a simple virtual machine..." New? Scheme interpreters are widely used? Quote "Every application includes an implementation of Lisp"
Lambda the Ultimate
Cite 19: Don't refer to unpublished papers.
"DSL implementation on top of Scheme VM." Sounds like a good idea. But why take (?) XPath?
The section "7. Sample: Python AST as XML". Too vague. Figure is desired. XML Python, Scheme.
The section "8. Sample: GNU find with XPath over the Filesystem". Report style: "we're done this and that". But a paper isn't a report.
Language independent implementation of DSL: