|
Hi Armin (or another expert interested in having XATTR_FILL regressions fixed), I spent a bit of time investigating a FillStyle bug (tdf#121111,
exposed by one of my fixes), but since this looks like a dangerous
area (with several long-standing regressions), I wanted to dig
deeper before attempting to push the fix I developed. The main
way I tried to dig deeper was to investigate all the other
FillStyle bugs. I've come up with fairly simple fixes for most of
them, but I'm afraid that my understanding of the situation hasn't
increased all that much. I've got a long way to go before I'll be
able to see a big enough picture on this. So I hope that with your
comprehensive understanding of the topic, you can help me close
out some of these bugs.
1.) https://gerrit.libreoffice.org/64185 tdf#121111 ww8export: fillstyle_NONE needs a nil background This one is probably more in the domain of a ww8export expert.
I'm pretty confident about this one.
2.) https://gerrit.libreoffice.org/#/c/64186/3 tdf#116071 ww8import: import to XATTR_FILL instead of RES_BACKGROUND I'm pretty confident that this one is on the right track, and with a cleaned up comment it probably could be submitted (after #1, so the unit test round-trips).
3.) https://gerrit.libreoffice.org/65609 tdf#99573 uiwriter: clear RES_BACKGROUND -> XATTR_FILL_* I'm pretty confident this one is on the right track, but there might be a better place to handle mapping a reset to RES_BACKGROUND to XATTR_FILL. Also, am I right in resetting ALL XATTR_FILL properties if RES_BACKGROUND is reset?
4.) https://gerrit.libreoffice.org/65811 tdf#115457 odfimport: FooterBackGraphic is not obsolete This is my fix from yesterday, and I'm not as confident in it especially since xmloff is still a new area for me. Likely marking FooterBackGraphic (formerly know as FooterBackGraphicURL) as obsolete was simply a mistake? But this has changed so much since 4.4 (FooterBackGraphic is "new" for LibreOffice 6.1) that I'm a bit lost.
5.) https://gerrit.libreoffice.org/64226 tdf#99125 sw: allow TextCursor to see FillProperties This one works nicely up until the clang job runs pythonTests, and identifies that the underlying infrastructure doesn't support it. [Assertion `!"unknown which - don't ask me for defaults"' failed]. Someone would need to carry me through this one if it completes with my name still attached to it.
Thanks, Justin
|