https://bugs.documentfoundation.org/show_bug.cgi?id=142264

Mike Kaganski <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|[email protected] |[email protected]
                   |desktop.org                 |

--- Comment #12 from Mike Kaganski <[email protected]> ---
https://gerrit.libreoffice.org/c/core/+/120688

This is not actually a true regression, just a pre-existing bug that was only
masked by a side effect of charts being incorrectly marked modified on open.

This is all about export filter not loading properly the OLE objects that may
get unloaded by the memory management in LO. The export reused the obsoleted
references to unloaded objects, which couldn't be exported correctly; and I
even suspect that existing "IsValidObject" testing chart objects validity in
sc/source/filter/xcl97/xcl97rec.cxx might be a hack to "workaround" this actual
filter bug.

I am thankful to paulystefan for this bug and the nice reproducers, that
eventually allowed me to notice the fixed number of object kept in the result.

I am suspecting now that there might be a similar problem elsewhere in our
code, that could be responsible for many "I lost images in a document"
complaints that were impossible to investigate, because there were never a
reliable reproducer. Now I suppose that adding a substantial number of
different images, while reducing GraphicMemoryLimit (or TotalCacheSize) might
allow to find a good reproducing scenario.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to