Ihor Radchenko <yanta...@posteo.net> writes:
>
> org-compile-file demanded the compilation target to be newer than .tex
> source, which is apparently too strong requirement.
>
> Fixed, on main.
> https://git.savannah.gnu.org/cgit/emacs/org-mode.git/commit/?id=499bd92e2
Looks like the new requirement too is too strong. Any latex error during
the process now causes export to apparently fail with the message:
File "testfile.pdf" wasn't produced
This happens even if the error is non-critical and a PDF /is/ in fact
produced. So the error message can be misleading.
To reproduce, try e.g. adding a bibtex step to org-latex-pdf-process and
exporting a minimal Org file without a bibliography to PDF.
(setq org-latex-pdf-process
'("%latex -interaction nonstopmode -output-directory %o %f"
"bibtex %f"
"%latex -interaction nonstopmode -output-directory %o %f"
"%latex -interaction nonstopmode -output-directory %o %f"))
Other legacy bugs I've found that the PDF export process has so far
silently forgiven, but now chokes on, include an options mismatch in the
url package and a character not found in the maths font.
Assuming that what I see is real and not a result of mixed installation
or something:
On the one hand, this has the merit of alerting me to long-standing,
non-critical flaws in my setup and documents. On the other, it's taken
time to get to the bottom of them, and I think we will see a lot of
confused users with documents that exported well enough to PDF but now
fail. (It may also be at the bottom of some new trouble I'm seeing with
png export of latex src blocks, but I'm not sure about that.)
If solving Jake's problem has led to even more false positives, perhaps
499bd92e should be reverted? Or is there a more discriminating test
we could use? If the change is kept, the error message should say the
PDF was produced with errors when that is what happens.
Regards,
Christian
