Hi Stéphane, > * when asserting an equality, e.g.: > CPPUNIT_ASSERT_EQUAL(OString("Im19"), rIDs[0]) > <https://opengrok.libreoffice.org/xref/core/vcl/qa/cppunit/pdfexport/pdfexport.cxx?r=4390f233&mo=95984&fi=2214#2268> > o what do strings like "Im19" or "Im12" mean in these contexts? There are comments above the assert, they are meant to verify that "we don't have 3 times the same page". You can also run 'git blame' on the relevant lines to see this was added in commit 4505bc1701be73aab6543520cd342f0f0c8b7853. > * what is the "W array" that is not found in: CPPUNIT_ASSERT(pWArray) > <https://opengrok.libreoffice.org/xref/core/vcl/qa/cppunit/pdfexport/pdfexport.cxx?r=4390f233&mo=95984&fi=2214#2661> It's a key in a PDF dictionary for a font object, did you attempt to search in the PDF spec to see what is the exact meaning? > * why could it be that rDocument.Read() fails at: > CPPUNIT_ASSERT(rDocument.Read(aStream)) > <https://opengrok.libreoffice.org/xref/core/vcl/qa/cppunit/pdfexport/pdfexport.cxx?r=4390f233&mo=95984&fi=2214#912> > (in the load function) That means that the pdf tokenizer was unable to create a list of PDF tokens from the input byte array. Usually there are some warnings printed (in debug builds) before that happens, pointing at some more specific problem. > Is there someone with more knowledge about tests and PDFs who could put me > on the right path? In general, running git blame on the code you don't understand is useful. You can also read comments around the code. Last, it seems your patch changes 2 settings at the same time: it would be good to change one at a time: it's always easier to debug a problem if you only changed one thing and you know the breakage is caused by exactly that change. > (I am not subscribed to the dev mailing list so please send a reply to me as > well as the list! 😁) It would great to subscribe though; that way only we can help you, but you can't help others. Regards, Miklos