[GSOC] Tiled Rendering, ahunt, Weekly Report 04 / WK24

[Andrzej Hunt]

Hi All,

Status for the last week:

* LOKit patches cleaned up/completed for integration.

* Fixed the remaining alpha channel issues (hardcoded to 0xFF) -- it 
turns out the
   svp/headless backend also does raw bitmap manipulation (and these can 
then be
   copied bit-for-bit into basebmp bitmaps if the formats "match"), so 
now no more
   artefacts.

* Rebased onto the updated master with libreofficekit (branch was 
previously based
   on the old liblibreoffice naming -- some pain around patch changes etc.).

* Moved most of the gtktiledviewer into a GTK+ widget (LibreOfficeKitGtk +
   lok_docview) -- gtktiledviewer is now a very thin wrapper around that 
widget.

* Fixed the "missing content" -- I discovered that the area in which 
text/images are
    rendered corresponds to what is shown in the main document window 
(visible
    when using the existing tiledrendering app, but hidden when using svp):
** PaintTile previously used MakeVisible(), but that only scrolls the 
content:
     if we try to paint a tile that is larger than the area in the 
window then
     content outside of the visible area will be missing.
** Instead we now set the tile area to be explicitly visible in the 
SwView. This is
     somewhat inelegant (and corrupts the content of the background 
window which
      is shown with the tilederendering app), but seems to work reliably 
(and for most
      uses of tiled rendering that window isn't wanted anyway so...).
*** (Could potentially reset the SwView area back after doing the 
rendering to prevent
       this? Might experiment with this this week)
** Longer term: not 100% sure what we want to do (ideally remove the whole
     concept of having to have a SalFrame associated with every document and
     decouple rendering from other assumptions etc.?).

* Tiny bit of refactoring to cleanup how we get access to tiled 
rendering to introduce
    a common interface which can be reused for Calc/Impress etc.

-- Andrzej