Reason:
Currently for chart prints metafile replacements are used (same as for all other
OLE objects).
For 3D shapes resolution dependent bitmaps are generated and contained in the
meta file.

I will try whether I can change the framework to ask the chart directly for a
higher resolution output for prints and not use the standard meta files in that
case.

There are more scenarios where the wrong resolution for printing hurts:
1) When 3D shapes are used
2) When transparency is used
3) When many many data points are used for the chart some may be skipped for
performance optimization when they happen to hit the same pixel. With the higher
printer resolution those points then can be missing.

Created attachment 49091 [details]
doc with example charts

Pay attention to print only the first page of the example.
When printing it you might see that some of the points are missing in the bar
chart. The 3D pie really looks ugly jagged. And in the line chart with
transparency the grid lines partly vanish (the latter was also reported in issue
41842).

Fixed in CWS chart15:
The transparency problems seem to have a different cause, so they are handled in
the separate issue 41842 and not here.
The existing chart ole replacement images are no longer used for printing or pdf
export. Instead the chart is now directly asked by the framework to produce a
high resolution output. Therefor the charts must be loaded if that has not
happened before. For 3D charts a higher resolution bitmap is created and send to
the printer. As the amount of data is bigger this takes more time now. Also when
for charts with many data points on higher resolution more data points can be
distinguished they are created now.

->Thomas, please verify in CWS chart15.

Thomas, please ensure that the performance for printing charts is still
acceptable. As said it is slower now because charts might be loaded and more
data needs to be transported in some cases. That is a necessary price for a
correct print.

For me the performance is still acceptable. -> verified

Seen ok in current master -> closed