In this case now inset dimensions are dynamic as it will depend on
position of inset.
Current implementation of dimensions has 3 members
{wid - width,
ascent - asc,
des-descent}.
With three box implementation, I think we will need four members
{ wid1-width of first row,
wid2-width of last row,
asc-ascent,
des-descent}.
Now to compute dimensions, I have thought something like below:
max_width (Maximum width of current viewport)
(x,y) (current position)
There will be 2 cases:
1) text of inset has only one row:
fig1:
[[
original inset text:
This is inset text.
with 3 box inset implementation:
This is the normal text.This is the normal text.[[This is inset text.]]This
is original text again.
]]
fig2:
[[
original inset text:
This is inset text.This is inset text.This is inset text.
with 3 box inset implementation:
This is the normal text.This is the normal text.[[This is inset text. This
is inset text.]]This is the normal text again.
]]
In this case, if (max_width - x) > width of row of original text
then
**fig1**
w1=width of row of original text
w2=0
else
**fig2**
w1= max_width - x
w2= width of row of original text - w1
2) text of inset has more than one rows:
fig3:
[[
Original inset Text
Foot This is the foonote which is pretty long, and what LyX would
normally do is put
this whole thing on a few lines by itself. What we want is for it
to flow with the
normal text.
with 3 box inset implementation:
This is the normal text. This is the normal text.[[Foot This is
the foonote which is
pretty long, and what LyX would normally do is put this whole
thing on a few lines
by itself. What we want is for it to flow with the normal text.]]
This is the normal
text again.
]]
fig4:
[[
Original inset text:
Foot This is the foonote which is pretty long, and what LyX would
normally do is put
this whole thing on a few lines by itself. What we want is for it
to flow with the
normal text.This is inset text. This is inset text. This is inset text.
with 3 box inset implementation:
This is the normal text. This is the normal text.[[Foot This is
the foonote which is
pretty long, and what LyX would normally do is put this whole
thing on a few lines
by itself. What we want is for it to flow with the normal
text.This is inset text. This
is inset text. This is inset text.]] This is the normal text again.
]]
In this case,
w1= max_width - x
if (max_width - x) > width of last row
of original text
then
**fig3**
w2= max_width + width of last
row of original text - w1
else
**fig4**
w2= width of last row of
original text - w1
and there will be n+1 rows in
inset if original text of inset has n rows
With this information and current position coordinates, polygon
covering inset text can be drawn easily.
Now with this implementation, I think I have to re-implement methods
like TextMetrics:drawParagraph, TextMetrics:redoParagraph ,methods in
rowPainter class and maybe some others because with current
implementation, text is painted row by row in original text and with
new implementation of 3 box model of insets, rows in original text is
not same as rows painted in workarea (like in fig3 and fig4) With
nested Insets complications may increase (I am not sure at this
moment) .
