Status report: consultancy team, week 5, 2024

Status reports
1

Caolan

  • consultancy meeting, calc stand up
  • confirm that new msan report of
    “use of uninitialized mem” is basically a false
    positive and figure out how to silence that
  • debug “double clicking in shape” resulting in a
    warning dialog for a little
  • continue with chart dark mode problems
  • tdf#151352 help out with tip of the day crash claimed
    as regression
  • Investigate calc text wider in render mode than in edit
    mode. Is it a kerning enabled/disabled in one vs the other?
    • no, different mapmodes
    • but same issue doesn’t seem too arise in desktop, why?
  • seeing lots of asserts when doing normal things
    • fix tdf#159386 itemset/itempool related regression
  • debug calc text width
    • spell checking mode is used in calc, and if the is a
      spelling error, even if the display spell checking is not enabled
      then an editengine is used to render the text. By default
      in desktop spell checking mode is not used, so that’s the
      initial difference. Mapmodes in desktop mode seem to match, and
      it might be that online cases don’t match.
  • more chart dark mode work
  • complete and merge dark mode chart problems
  • std::rng::entropy problem, review fix, back port.
  • can clearly reproduce in ci as 0, but locally
    std::rng::entropy give 32. No idea if the difference
    lies in version of libstdc++ or elsewhere.
  • usual perf trace for writer and calc
    • hot fix for apparent ScTableInfo over allocation based on
      that.
  • backport std::rng::entropy fixes to stable branches

Mike

Worked on:

— Inconsistent behaviour with document repair mechanism
— Bookmarks placed directly behind each other produce une[…]

Tomaz

Calc: add currency dropdown to the notebookbar

  • Backporting into master
    • reverted the MRU code in master
    • tried to keep some things from the MRU as they seem useful, but didn’t like how it was written
    • better to refactor from scratch, but this is an exercise for another time
  • Adapted to SfxItemSet changes
    • need to explicitly set which item can get ItemSurrogates

Impress: Font size changes in textbox when editing PPTX

  • Reproduced this particular issue
    • document (1 slide only) loads a too long time, but this is an independent issue - however annoying when debugging and trying out
  • Checked when and with what numbers we scale
    • Issue happens in SdrTextObj::onEditOutlinerStatusEvent
    • We re-fit the text and the rOutliner.CalcTextSizeNTP() doesn’t return a expected value, which influences the scaling
    • If we just don’t recalculate scaling in this case helped, but this recalculation is there for a reason
    • Trying to figure out why the difference in CalcTextSizeNTP()
  • Looking at other issues as well
  • Checking also if we can calculate the scaling inside the EditEng itself
    • Would be more consistent and probably faster
    • just loop CreateLines and modify parameters - font size changes by ± 1pt

Miklos

Worked on: