David Masterson <dsmaster...@icloud.com> writes: > Ihor Radchenko <yanta...@posteo.net> writes: > >> David Masterson <dsmaster...@icloud.com> writes: >> >>> org-publish publishes a lot of files for me via multiple backends. The >>> problem is that, in ox-latex's post scan of Latex output for warnings to >>> add to the *Warnings* buffer, it doesn't print out the filename >>> associated with the warning which makes it difficult to backtrack. >>> >>> My wish: >>> >>> I can (mostly) read Elisp, but ox-latex is somewhat more than I can >>> figure out. I'm hoping someone with more Lisp knowledge than me could >>> take a look. Is the starting place org-latex-compile--postprocess ? >>> I'm not sure if outfile would be the filename. >> >> The starting place is `org-compile-file'. Its LOG-BUF argument defines >> where the compilation output is dumped. You can augment that output >> with some information about which file is being compiled and using >> which commands. > > Hmm. Am I right in assuming LOG-BUF will be the *Messages* buffer? > Also, will ERR-MSG be the *Warnings* buffer? Then, it looks like the > appropriate thing for what I want would be to set 'ERR-MSG=LOG-BUF', but > only when publishing files. Unfortunately, I don't see a good way of > doing without some significant rewrite the publishing process. Is there > a useful hook that I'm not seeing?
Minor update: Unfortunately, I don't see a good way of doing this without some significant rewrite of the publishing process *unless* there is a global variable that's true when org-publish has been invoked. > Assumptions: > 1. Hand org-export-* is done to export a single file > 2. Hand org-publish is done to export many files (plus other stuff) > 3. It's OK (eh?) to have a separate *Warnings* buffer in #1 > 4. In #2, publish says which file it's working on in *Messages* buffer > 5. Interleaving #3 and #4 produces a good log where warnings can be > traced to the file that caused them > > Note: running org-publish via 'emacs --batch' (which I often do via a > Makefile) produces #5 just fine ...because warnings and messages are mixed in the output. -- David Masterson
