ideas for xslt while #37c3

While at congress, my thoughts are on things like how to form this very XML here. I had the idea a while ago: the id tag anchors to become the filenames, the title and published/updated to become three fixed lines and the extensions to determine the type attribute of content: text, xhtml, or the likes of markdown which can be turned into xhtml, or html which can be more or less turned into xhtml. Or a .summary.*, with a fourth fixed line for the link[alternate=]. --- And at first I have been thinking: just some bash or python. But now I discovered lots can be done with just... XSLT, my beloved. Although 2.0. It has the document() thing and the collection() ([ns]=fn:) thing. And there is the xsl:result-document with xsl:copy-of - per https://stackoverflow.com/questions/4833361/how-to-split-xml-file-into-many-xml-files-using-xslt/4833799#4833799 . These can be used to formulate multiple documents and to merge them - but just one thing that would be neat is lacking: It is only to be the 2017 XSLT 3.0 / XPath 3.1 that is introducing... a `sort` https://www.w3.org/TR/xslt-30/ , which i feel like is surprisingly hard to do without when managing na ordered collection of entries - unless we rely on some UB of directory traversal. Of course the solution is to supplement an quick-and-dirty indexing solution as part of a script. But still I had quite a moment of realizing XSLT can do for quite an... SSG, as a meta-engine at least.