Please ignore the noise. I had added the odt template to the file to be exported; it included the following line:
#+ODT_STYLES_FILE: Which I guess overrode my setting for odt_styles_dir. solved. On Mon, Jan 2, 2017 at 8:59 PM, Matt Price <mopto...@gmail.com> wrote: > I tried and then reverted a very simple change ot ox-odt.el -- I wanted to > export a unicode symbol instead of [ ] for an empty checkbox. > > After this, odt export fails mysteriously with the following: > > ------------ > Wrote /tmp/odt-3348W1J/meta.xml > Using vacuous schema > OpenDocument export failed: Cannot revert nonexistent file > /tmp/odt-3348W1J/styles.xml > Result: "OpenDocument export failed: Cannot revert nonexistent file > /tmp/odt-3348W1J/styles.xml" > ------------ > > I tried instrumenting org-odt-export-to-odt but was not able to narrow the > error down further than: > > (org-odt--export-wrap > outfile > (let* ((org-odt-embedded-images-count 0) > (org-odt-embedded-formulas-count 0) > (org-odt-automatic-styles nil) > (org-odt-object-counters nil) > ;; Let `htmlfontify' know that we are interested in collecting > ;; styles. > (hfy-user-sheet-assoc nil)) > ;; Initialize content.xml and kick-off the export process. > (let ((output (org-export-as 'odt subtreep visible-only nil > ext-plist)) > (out-buf (progn > (require 'nxml-mode) > (let ((nxml-auto-insert-xml-declaration-flag nil)) > (find-file-noselect > (concat org-odt-zip-dir "content.xml") t))))) > (with-current-buffer out-buf (erase-buffer) (insert output))))) > > > i don't seem to be able to instrument the org-odt--export-wrap macro, but > maybe the error report comes from this section around line 4067: > > (error > ;; Cleanup work directory and work files. > (funcall --cleanup-xml-buffers) > (message "OpenDocument export failed: %s" > (error-message-string err) > > This is with a pretty recent git master w/ a few changes applied only to > ox-html.el. > > ~git diff master~ > > shows no changes to ox-odt.el, but I restarted emacs and restarted my > laptop just in case -- the problem still persists. > > Thanks! > > Matt > > >
