Hi Regina, On Fri, Apr 16, 2021 at 12:42:28AM +0200, Regina Henschel <rb.henschel@xxxxxxxxxxx> wrote: > Now I stuck, because some unit tests fail with small differences. But before > I put a lot of time into it, to find whether the values in the unit tests > are wrong or whether and where my patch is incorrect, I would like, if you > have a look at the patch in general. The approach looks reasonable to me in general. As I understand it, the LO group shape is a simple container of shapes (with a name), while the MSO group shape can have its own rotation, scaling, etc -- so their document model can be more complex. Based on this, it's expected that you may need the "original" rotation of the group shape in writerfilter/. Regarding the unit tests: I try to always document how a test failed without the original fix. So in case you see that e.g. now the test fails because of a 98 vs 99 difference and the comment says the old value was 42, then this may suggest that only the test was poor and it can be adapted. You can always manually compare the import result of a particular test document with and without your fix: if you see no differences, then that also suggests that you are OK to change the test. Otherwise it's likely that the test caught some badness and you need to improve your patch. Hope that helps, Miklos _______________________________________________ LibreOffice mailing list LibreOffice@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/libreoffice