[++kendy]
On 12/17/19 2:18 AM, Stephan Bergmann wrote:
On 17/12/2019 01:46, Ashod Nakashian wrote:
On 12/16/19 11:59 AM, Stephan Bergmann wrote:
What's the reason for this change, and/or are there plans to enable
the tests again? (`git log -SIMPORT_PDF_ELEMENTS` shows this to be
the only commit ever mentioning that identifier, so the tests are
indeed disabled since then.)
The tests are conditional on IMPORT_PDF_ELEMENTS because with the
PDFium import logic, PDFs are imported as images, not individual
elements (users can break the image into editable elements from the UI).
Since these tests rely on importing PDF elements individually, the
PDFium importing logic breaks them.
But where does the the IMPORT_PDF_ELEMENTS macro get defined, if
anywhere?
Either that patch is missing, or I never got around to implementing it,
since it would remain unused (i.e. no build would define it).
As I said, the idea here is to import PDFs as images using PDFium, which
has a very high quality (and is much faster) than importing individual
editable elements. So that's the "new" way of importing.
The user then breaks the image to editable elements (which is not
perfect, but very close to the old way). This also removes the need to
have out-of-process PDF parsing (due to license conflicts).
PDFium should overall be better, and where it isn't, we just need to
improve it. As far as I can tell, the old importing logic gives us not
much advantage.
So these tests aren't really useful anymore. They should be modified to
do import+break and then check the results. But that isn't
straight-forward and I ran out of time to do it.
Let me know if you have any concerns please.
Thanks.
_______________________________________________
LibreOffice mailing list
LibreOffice@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/libreoffice
