On Mon, Dec 01, 2003 at 01:18:05PM +0000, Angus Leeming wrote:
> Index: buffer.h
> - /** The list of paragraphs.
> + /** The list of paragraphs().
> This is a linked list of paragraph, this list holds the
> whole contents of the document.
> */
Grmpf. As soon as I introduce some verbosity people start to complain...
> Index: text2.C
> - the beginning of paragraphs. This happens if you have
> + the beginning of paragraphs(). This happens if you have
>
>
> Thought: do you still need to forward-declare BufferView now?
> Index: iterators.h
> @@ -20,6 +20,7 @@
> class LyXText;
> class InsetOld;
> class Cursor;
> +class Buffer;
> class BufferView;
> class PosIterator;
>
> What has happended to this:
> Index: lyxfunc.C
> - case LFUN_INSET_TOGGLE: {
> - LyXText * lt = view()->getLyXText();
> - disable = !(isEditableInset(lt->getInset())
> - || (lt->inset_owner
> - && lt->inset_owner->owner()
> - && lt->inset_owner->owner()->isOpen()));
> - break;
> - }
>
> ... Ahhhhhhhhh. Neat (I see the code in insetcollapsable).
Most of the other inset->owner() usages are probably removable like
this. That's why my suspicion that InsetOld::owner_ is not needed at
all.
> Thought: Buffer now contains a LyxText variable that contains the
> entire document, right?
Hm, yes.
> Is 'Buffer" becoming redundant, given that most of its member
> functions could/should be implemented as free functions?
Maybe. Note that a LyXText is 'nestable', a Buffer is not. So the
main text plays two roles. Having them separated in 'Buffer' (read/write
files etc) and 'LyXText' (basic editing) does not sound wrong.
> This makes me feel uneasy, maybe because I never really got to grips
> with the 'inherit' stuff. Is the change really Ok?
>
> @@ -82,13 +71,7 @@ void InsetOld::setBackgroundColor(LColor
>
> LColor_color InsetOld::backgroundColor() const
> {
> - if (background_color_ == LColor::inherit) {
> - if (owner())
> - return owner()->backgroundColor();
> - else
> - return LColor::background;
> - } else
> - return LColor::color(background_color_);
> + return LColor::color(background_color_);
> }
I've played a bit around and it got e.g. the Note inset right. I was
actually expecting the need to pass the 'active' background color around
using the PainterInfo struct but so far I've not yet seen a case where
this is needed.
> Ditto. Really Ok?
> Index: insets/inset.h
> - /// check if the font of the char we want inserting is correct
> - /// and modify it if it is not.
> - virtual bool checkInsertChar(LyXFont &);
>
> - /* needed for widths which are % of something
> - returns the value of \textwidth in this inset. Most of the
> - time this is the width of the workarea, but if there is a
> - minipage somewhere, it will be the width of this minipage */
> - virtual int latexTextWidth(BufferView *) const;
I think so. This pretty much was dead code and the only suspicious case
'minipage' gets it right without this.
> I think that these showInsetDialog member functions could safely be
> discarded, given tht many insets implement this directly in
> priv_dispatch:
>
> Index: insets/insetwrap.h
> - bool showInsetDialog(BufferView *) const;
> - ///
> - int latexTextWidth(BufferView *) const;
> + bool showInsetDialog(BufferView *) const;
>
> Man! You've been busy!
It was the third or fourth attempt on the matter, so it wasn't that
difficult...
Andre'
