Spreadsheet FR

Spreadsheet FR

[Russell Adams]

Fellow Org'ers,

I adore the text spreadsheet, however there's one feature Excel
provides which I don't have in org.

I often use Excel for "lists", where I can sort or narrow the data by
specific criteria from a larger list.

Would it be feasible to "narrow" a table by criteria on a specific
field in between separators? Ie: only display those cells in field A
if they are > 2, or if field B matches "Pick Me!".

A nice feature would be updating the totals at the bottom with only
the visible data.

Just like the outline folding the goal would be to hide entries that
don't match, they should still remain.

So for example:

|--------+--------|
| *Lime* | *Cost* |
|--------+--------|
| Y      |      1 |
| Y      |      2 |
| Y      |      2 |
| N      |      3 |
| N      |      4 |
| Y      |      5 |
|--------+--------|
| Total  |     17 |
|--------+--------|
#+TBLFM: @8$2=vsum(@-I..@-II)

I can already sort by Lime or Cost, but if I filtered where Lime =
"Y", I would have:

|--------+--------|
| *Lime* | *Cost* |
|--------+--------|
| Y      |      1 |
| Y      |      2 |
| Y      |      2 |...
| Y      |      5 |
|--------+--------|
| Total  |     10 |
|--------+--------|
#+TBLFM: @8$2=vsum(@-I..@-II)

No loss of the lines, on reload or changing the view they would come
back, but note that the formula at the end didn't consider
them. Ideally the cell references wouldn't change (the example above
would break if recalc was hit).

I'm not sure whether this would be cumulative or whether the criteria
specification would need to allow for multiple logical conditions.

Would anyone else consider this a useful feature? I'm not sure how
difficult the implementation would be.

Thanks.

Re: Spreadsheet FR

[Carsten Dominik]

On Apr 1, 2010, at 11:46 PM, Russell Adams wrote:

> Fellow Org'ers,
>
> I adore the text spreadsheet, however there's one feature Excel
> provides which I don't have in org.
>
> I often use Excel for "lists", where I can sort or narrow the data by
> specific criteria from a larger list.
>
> Would it be feasible to "narrow" a table by criteria on a specific
> field in between separators? Ie: only display those cells in field A
> if they are > 2, or if field B matches "Pick Me!".

This one might be possible - but dangerous for losing data.

>
> A nice feature would be updating the totals at the bottom with only
> the visible data.

I don't think Excel works like this, does it?

>
> Just like the outline folding the goal would be to hide entries that
> don't match, they should still remain.
>
> So for example:
>
> |--------+--------|
> | *Lime* | *Cost* |
> |--------+--------|
> | Y      |      1 |
> | Y      |      2 |
> | Y      |      2 |
> | N      |      3 |
> | N      |      4 |
> | Y      |      5 |
> |--------+--------|
> | Total  |     17 |
> |--------+--------|
> #+TBLFM: @8$2=vsum(@-I..@-II)
>
> I can already sort by Lime or Cost, but if I filtered where Lime =
> "Y", I would have:
>
> |--------+--------|
> | *Lime* | *Cost* |
> |--------+--------|
> | Y      |      1 |
> | Y      |      2 |
> | Y      |      2 |...
> | Y      |      5 |
> |--------+--------|
> | Total  |     10 |
> |--------+--------|
> #+TBLFM: @8$2=vsum(@-I..@-II)
>
> No loss of the lines, on reload or changing the view they would come
> back, but note that the formula at the end didn't consider
> them. Ideally the cell references wouldn't change (the example above
> would break if recalc was hit).
>
> I'm not sure whether this would be cumulative or whether the criteria
> specification would need to allow for multiple logical conditions.
>
> Would anyone else consider this a useful feature? I'm not sure how
> difficult the implementation would be.
>
> Thanks.
>
>
> _______________________________________________
> Emacs-orgmode mailing list
> Please use `Reply All' to send replies to the list.
> Emacs-orgmode@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-orgmode

- Carsten

Re: Spreadsheet FR

[Russell Adams]

On Fri, Apr 02, 2010 at 09:49:46AM +0200, Carsten Dominik wrote:
>
>> Would it be feasible to "narrow" a table by criteria on a specific
>> field in between separators? Ie: only display those cells in field A
>> if they are > 2, or if field B matches "Pick Me!".
>
> This one might be possible - but dangerous for losing data.

Ideally we're just narrowing away lines, collapsing a table like org
collapses headlines.

Why would that be more likely to cause data loss?

>> A nice feature would be updating the totals at the bottom with only
>> the visible data.
>
> I don't think Excel works like this, does it?

Absolutely. Excel 2003 and up support "lists" or in 2007 "tables",
where you can do filter and summary work. I use it all the time for
customer planning because its the closest thing to grep/awk for Windows.

I think the real trick here is making the formulas ignore narrowed content.

If you'd like I can send you a sample.

Thanks.


------------------------------------------------------------------
Russell Adams                            RLAdams@AdamsInfoServ.com

PGP Key ID:     0x1160DCB3           http://www.adamsinfoserv.com/

Fingerprint:    1723 D8CA 4280 1EC9 557F  66E8 1154 E018 1160 DCB3

Re: Spreadsheet FR

[Darlan Cavalcante Moreira]

How can you expand/narrow the disabled lines if you want to show/hide them?
This will probably interfere with normal TAB use inside a table.

I think that the ability to "disable" some lines based on data in a column,
such as in your example, can be very useful. But maybe we don't need to
really collapse the lines. Only changing the face could be enough.

- Darlan

At Fri, 2 Apr 2010 03:16:58 -0500,
Russell Adams <RLAdams@AdamsInfoServ.Com> wrote:
> 
> On Fri, Apr 02, 2010 at 09:49:46AM +0200, Carsten Dominik wrote:
> >
> >> Would it be feasible to "narrow" a table by criteria on a specific
> >> field in between separators? Ie: only display those cells in field A
> >> if they are > 2, or if field B matches "Pick Me!".
> >
> > This one might be possible - but dangerous for losing data.
> 
> Ideally we're just narrowing away lines, collapsing a table like org
> collapses headlines.
> 
> Why would that be more likely to cause data loss?
> 
> >> A nice feature would be updating the totals at the bottom with only
> >> the visible data.
> >
> > I don't think Excel works like this, does it?
> 
> Absolutely. Excel 2003 and up support "lists" or in 2007 "tables",
> where you can do filter and summary work. I use it all the time for
> customer planning because its the closest thing to grep/awk for Windows.
> 
> I think the real trick here is making the formulas ignore narrowed content.
> 
> If you'd like I can send you a sample.
> 
> Thanks.
> 
> 
> ------------------------------------------------------------------
> Russell Adams                            RLAdams@AdamsInfoServ.com
> 
> PGP Key ID:     0x1160DCB3           http://www.adamsinfoserv.com/
> 
> Fingerprint:    1723 D8CA 4280 1EC9 557F  66E8 1154 E018 1160 DCB3
> 
> 
> _______________________________________________
> Emacs-orgmode mailing list
> Please use `Reply All' to send replies to the list.
> Emacs-orgmode@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-orgmode

Re: Spreadsheet FR

[Matt Lundin]

Hi Russell,

Russell Adams <rladams@adamsinfoserv.com> writes:

> I adore the text spreadsheet, however there's one feature Excel
> provides which I don't have in org.
>
> I often use Excel for "lists", where I can sort or narrow the data by
> specific criteria from a larger list.
>
> Would it be feasible to "narrow" a table by criteria on a specific
> field in between separators? Ie: only display those cells in field A
> if they are > 2, or if field B matches "Pick Me!".

One way to achieve this behavior is by using properties + Eric Schulte's
org-collector.el in contrib. You can enter the data in a subtree using
properties and columns and then capture a subset of that data in a table
using a dynamic block. E.g., given the following subtree...

--8<---------------cut here---------------start------------->8---
* My shopping list
  :PROPERTIES:
  :COLUMNS:  %40ITEM %10AMOUNT %10TYPE
  :ID:       41101c80-85b9-4af6-8138-8ad460a2949e
  :END:
** Bananas
   :PROPERTIES:
   :AMOUNT:   2.65
   :TYPE:     fruit
   :END:
** Cheese
   :PROPERTIES:
   :AMOUNT:   4.54
   :TYPE:     dairy
   :END:
** Flour
   :PROPERTIES:
   :AMOUNT:   3.00
   :TYPE:     grain
   :END:
** Dish soap
   :PROPERTIES:
   :AMOUNT:   1.29
   :TYPE:     kitchen
   :END:
** Lettuce
   :PROPERTIES:
   :AMOUNT:   1.80
   :TYPE:     vegetable
   :END:
** Carrots
   :PROPERTIES:
   :AMOUNT:   2.56
   :TYPE:     vegetable
   :END:
** Milk
   :PROPERTIES:
   :AMOUNT:   3.25
   :TYPE:     dairy
   :END:
--8<---------------cut here---------------end--------------->8---

...you can use org-collector to capture a subset of data as follows...

--8<---------------cut here---------------start------------->8---
* Data

#+begin: propview :id "41101c80-85b9-4af6-8138-8ad460a2949e" :conds ((string= TYPE "vegetable")) :cols (ITEM AMOUNT TYPE)
| "ITEM"    | "AMOUNT" | "TYPE"      |
|-----------+----------+-------------|
| "Lettuce" |      1.8 | "vegetable" |
| "Carrots" |     2.56 | "vegetable" |
|-----------+----------+-------------|
|           |     4.36 |             |
#+TBLFM: $LR2=vsum(@I+1..@II-1)
#+end
--8<---------------cut here---------------end--------------->8---

Change the conditions to match dairy items greater than 4 and you get
the following:

--8<---------------cut here---------------start------------->8---
* Data

#+begin: propview :id "41101c80-85b9-4af6-8138-8ad460a2949e" :conds ((and (string= TYPE "dairy") (> AMOUNT 4))) :cols (ITEM AMOUNT TYPE)
| "ITEM"   | "AMOUNT" | "TYPE"  |
|----------+----------+---------|
| "Cheese" |     4.54 | "dairy" |
|----------+----------+---------|
|          |     4.54 |         |
#+TBLFM: $LR2=vsum(@I+1..@II-1)
#+end
--8<---------------cut here---------------end--------------->8---

Best,
Matt

Re: Spreadsheet FR

[Dan Davison]

Hi Russell,

Carsten Dominik <carsten.dominik@gmail.com> writes:

> On Apr 1, 2010, at 11:46 PM, Russell Adams wrote:
>
>> Fellow Org'ers,
>>
>> I adore the text spreadsheet, however there's one feature Excel
>> provides which I don't have in org.
>>
>> I often use Excel for "lists", where I can sort or narrow the data by
>> specific criteria from a larger list.
>>
>> Would it be feasible to "narrow" a table by criteria on a specific
>> field in between separators? Ie: only display those cells in field A
>> if they are > 2, or if field B matches "Pick Me!".

How about keeping a master table containing all the information, and
then generating narrowed views as separate tables? The babel way to do
this would be to have a block function "filter-table" (provided below)
and then call it where needed:

#+TBLNAME: lime-table
| *Lime* | *Cost* |
|--------+--------|
| Y      |      1 |
| Y      |      2 |
| Y      |      2 |
| N      |      3 |
| N      |      4 |
| Y      |      5 |
| Total  |     17 |
#+TBLFM: @8$2=vsum(@-I..@-II)

#+call: filter-table(table=lime-table, field=0, value="Y")

#+results: filter-table(table=lime-table, field=0, value="Y")
| *Lime* | *Cost* |
|--------+--------|
| Y      |      1 |
| Y      |      2 |
| Y      |      2 |
| Y      |      5 |

As you can see I got rid of a few horizontal separator lines, and we
don't currently have totals. I'll suggest fixes for that below, but my
main point is that although it may make sense to extend org-mode, it's
already easy to do this in org-babel.

The separator lines are details, we can fix that by tinkering (if
there's interest we could continue this thread to do so). As for the
totals, we could go two routes:

1. We could add a final total row and use a table formula to compute the
   total
2. We could add the total row and compute the totals in a block
   function.

As far as I know there's no way to add a final row using a table formula
-- is that right? But we could automate (1) by wrapping the call to
filter-table in a second block function that is specific to this
problem:

#+call: filter-with-total-line(table=lime-table, field=0, value="Y")

#+results: filter-with-total-line(table=lime-table, field=0, value="Y")
| *Lime* | *Cost* |
|--------+--------|
| Y      |      1 |
| Y      |      2 |
| Y      |      2 |
| Y      |      5 |
| Total  |     10 |
#+TBLFM: @6$2=vsum(@-I..)

OK, so I hard wired a 6 into the TBLFM line. Perhaps someone can tell me
how to refer to the last line of the table. Also, it might be nice if
babel could apply the table formula automaticaly after generating the
table.

Here are the block functions. Once someone has written them, nobody else
needs to know anything about the implementation. They can be placed in
the same file (or in a different file, and added to your library of
babel using org-babel-lob-ingest). I've already added filter-table to
LoB on Worg.

#+function: filter-table(table, field, value)
#+begin_src emacs-lisp
  (defun org-lob-filter-table (table field value)
    (if (and (> (length table) 1)
             (eq (second table) 'hline))
        (append
         (list
          (first table)
          'hline)
         (org-lob-filter-table (cddr table) field value))
      (delq nil
            (mapcar (lambda (row)
                      (cond
                       ((eq row 'hline) 'hline)
                       ((equal (nth field row) value) row)))
                    table))))
  
  (org-lob-filter-table table field value)
#+end_src

#+function: filter-with-total-line(table, field, value)
#+begin_src emacs-lisp
  (append (org-lob-filter-table table field value)
          '(("Total" "")))
#+end_src

For those who are still reading, 

1. These are written in elisp so that they're usable by anyone, although
   they would be one-liners in R.
2. The second one uses a function defined in the first one. This may be
   OK, but a perhaps preferable solution would be

#+function: append-total-line(table)
#+begin_src emacs-lisp
  (append table '(("Total" "")))
#+end_src

#+call: append-total-line(table=filter-table(table=lime-table, field=0, value="Y"))

Dan


>
> This one might be possible - but dangerous for losing data.
>
>>
>> A nice feature would be updating the totals at the bottom with only
>> the visible data.
>
> I don't think Excel works like this, does it?
>
>>
>> Just like the outline folding the goal would be to hide entries that
>> don't match, they should still remain.
>>
>> So for example:
>>
>> |--------+--------|
>> | *Lime* | *Cost* |
>> |--------+--------|
>> | Y      |      1 |
>> | Y      |      2 |
>> | Y      |      2 |
>> | N      |      3 |
>> | N      |      4 |
>> | Y      |      5 |
>> |--------+--------|
>> | Total  |     17 |
>> |--------+--------|
>> #+TBLFM: @8$2=vsum(@-I..@-II)
>>
>> I can already sort by Lime or Cost, but if I filtered where Lime =
>> "Y", I would have:
>>
>> |--------+--------|
>> | *Lime* | *Cost* |
>> |--------+--------|
>> | Y      |      1 |
>> | Y      |      2 |
>> | Y      |      2 |...
>> | Y      |      5 |
>> |--------+--------|
>> | Total  |     10 |
>> |--------+--------|
>> #+TBLFM: @8$2=vsum(@-I..@-II)
>>
>> No loss of the lines, on reload or changing the view they would come
>> back, but note that the formula at the end didn't consider
>> them. Ideally the cell references wouldn't change (the example above
>> would break if recalc was hit).
>>
>> I'm not sure whether this would be cumulative or whether the criteria
>> specification would need to allow for multiple logical conditions.
>>
>> Would anyone else consider this a useful feature? I'm not sure how
>> difficult the implementation would be.
>>
>> Thanks.
>>
>>
>> _______________________________________________
>> Emacs-orgmode mailing list
>> Please use `Reply All' to send replies to the list.
>> Emacs-orgmode@gnu.org
>> http://lists.gnu.org/mailman/listinfo/emacs-orgmode
>
> - Carsten
>
>
>
>
>
> _______________________________________________
> Emacs-orgmode mailing list
> Please use `Reply All' to send replies to the list.
> Emacs-orgmode@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-orgmode

Re: Spreadsheet FR

[Russell Adams]

On Fri, Apr 02, 2010 at 10:44:23AM -0400, Dan Davison wrote:
> How about keeping a master table containing all the information, and
> then generating narrowed views as separate tables? The babel way to do
> this would be to have a block function "filter-table" (provided below)
> and then call it where needed:
> 
> #+TBLNAME: lime-table
> | *Lime* | *Cost* |
> |--------+--------|
> | Y      |      1 |
> | Y      |      2 |
> | Y      |      2 |
> | N      |      3 |
> | N      |      4 |
> | Y      |      5 |
> | Total  |     17 |
> #+TBLFM: @8$2=vsum(@-I..@-II)
> 
> #+call: filter-table(table=lime-table, field=0, value="Y")
> 
> #+results: filter-table(table=lime-table, field=0, value="Y")
> | *Lime* | *Cost* |
> |--------+--------|
> | Y      |      1 |
> | Y      |      2 |
> | Y      |      2 |
> | Y      |      5 |
> 
> As you can see I got rid of a few horizontal separator lines, and we
> don't currently have totals. I'll suggest fixes for that below, but my
> main point is that although it may make sense to extend org-mode, it's
> already easy to do this in org-babel.

That's a better way to do it than properties in my opinion.

Whether extension was required or not, my only point was that for
small lists the only reason I still use OO or Excel was for the
ability to dynamically narrow a list.

Given org narrows outlines, I thought it might fit.

I'll have to play with babel now!

> OK, so I hard wired a 6 into the TBLFM line. Perhaps someone can tell me
> how to refer to the last line of the table. Also, it might be nice if
> babel could apply the table formula automaticaly after generating the
> table.

This is why I use @-I..@-II, the separators are your cues. There's an
LR symbol, but I don't use it often.

That alone is a good reason to make sure separator lines are preserved.

Thanks.

------------------------------------------------------------------
Russell Adams                            RLAdams@AdamsInfoServ.com

PGP Key ID:     0x1160DCB3           http://www.adamsinfoserv.com/

Fingerprint:    1723 D8CA 4280 1EC9 557F  66E8 1154 E018 1160 DCB3

Re: Spreadsheet FR

[Russell Adams]

On Fri, Apr 02, 2010 at 09:55:16AM -0500, Russell Adams wrote:
> > As you can see I got rid of a few horizontal separator lines, and we
> > don't currently have totals. I'll suggest fixes for that below, but my
> > main point is that although it may make sense to extend org-mode, it's
> > already easy to do this in org-babel.
> 
> That's a better way to do it than properties in my opinion.
> 
> Whether extension was required or not, my only point was that for
> small lists the only reason I still use OO or Excel was for the
> ability to dynamically narrow a list.
> 
> Given org narrows outlines, I thought it might fit.

One issue I just realized is that babel in this format will strictly
be a *viewer*, that you could not reduce the dataset and edit it in
adjacent lines.



------------------------------------------------------------------
Russell Adams                            RLAdams@AdamsInfoServ.com

PGP Key ID:     0x1160DCB3           http://www.adamsinfoserv.com/

Fingerprint:    1723 D8CA 4280 1EC9 557F  66E8 1154 E018 1160 DCB3

Re: Spreadsheet FR

[Matt Lundin]

Dan Davison <davison@stats.ox.ac.uk> writes:

> Hi Russell,
>
> Carsten Dominik <carsten.dominik@gmail.com> writes:
>
>> On Apr 1, 2010, at 11:46 PM, Russell Adams wrote:
>>
>>> Fellow Org'ers,
>>>
>>> I adore the text spreadsheet, however there's one feature Excel
>>> provides which I don't have in org.
>>>
>>> I often use Excel for "lists", where I can sort or narrow the data by
>>> specific criteria from a larger list.
>>>
>>> Would it be feasible to "narrow" a table by criteria on a specific
>>> field in between separators? Ie: only display those cells in field A
>>> if they are > 2, or if field B matches "Pick Me!".
>
> How about keeping a master table containing all the information, and
> then generating narrowed views as separate tables? The babel way to do
> this would be to have a block function "filter-table" (provided below)
> and then call it where needed:

This is fantastic. Thanks, Dan!

- Matt

Re: Spreadsheet FR

[Carsten Dominik]

On Apr 2, 2010, at 10:16 AM, Russell Adams wrote:

> On Fri, Apr 02, 2010 at 09:49:46AM +0200, Carsten Dominik wrote:
>>
>>> Would it be feasible to "narrow" a table by criteria on a specific
>>> field in between separators? Ie: only display those cells in field A
>>> if they are > 2, or if field B matches "Pick Me!".
>>
>> This one might be possible - but dangerous for losing data.
>
> Ideally we're just narrowing away lines, collapsing a table like org
> collapses headlines.
>
> Why would that be more likely to cause data loss?

Because there would be ellipsis in unfamiliar places and the danger to  
accidentally modify invisible text would be higer.

>
>>> A nice feature would be updating the totals at the bottom with only
>>> the visible data.
>>
>> I don't think Excel works like this, does it?
>
> Absolutely. Excel 2003 and up support "lists" or in 2007 "tables",
> where you can do filter and summary work. I use it all the time for
> customer planning because its the closest thing to grep/awk for  
> Windows.

OK, I did not know this.

>
> I think the real trick here is making the formulas ignore narrowed  
> content.
>
> If you'd like I can send you a sample.

Please do - but that does not mean that I am going to implement this.   
This would be tricky.

- Carsten

Re: Spreadsheet FR

[Xiao-Yong Jin]

On Sun, 4 Apr 2010 09:01:30 +0200, Carsten Dominik wrote:

> On Apr 2, 2010, at 10:16 AM, Russell Adams wrote:

>> On Fri, Apr 02, 2010 at 09:49:46AM +0200, Carsten Dominik wrote:
>>> 
>>>> Would it be feasible to "narrow" a table by criteria on a specific
>>>> field in between separators? Ie: only display those cells in field A
>>>> if they are > 2, or if field B matches "Pick Me!".
>>> 
>>> This one might be possible - but dangerous for losing data.
>> 
>> Ideally we're just narrowing away lines, collapsing a table like org
>> collapses headlines.
>> 
>> Why would that be more likely to cause data loss?

> Because there would be ellipsis in unfamiliar places and the danger to
> accidentally modify invisible text would be higer.

I have an idea.  We can leave the original table intact,
while making a new dynamically generated table derived from
the original one.

#+tblname: myexptable
| x | y |            z |
|---+---+--------------|
| 1 | 1 |   0.36787944 |
| 1 | 2 |   0.13533528 |
| 2 | 3 |  0.099574137 |
| 2 | 4 |  0.036631278 |
| 3 | 5 |  0.020213841 |
| 3 | 6 | 7.4362565e-3 |
#+TBLFM: $3=$1*exp(-$2)

#+DERIVEDTBL :origin myexptable :criteria $1==2
| x | y |           z |
|---+---+-------------|
| 2 | 3 | 0.099574137 |
| 2 | 4 | 0.036631278 |

It is just a quick thought.  I am not sure how easy it is to
implement it, but at least it is unlikely to make any
confusion to users.

                                        -- jxy
-- 
J    c/*    __o/*
X    <\     * (__
Y    */\      <

Re: Spreadsheet FR

[Russell Adams]

On Sun, Apr 04, 2010 at 12:46:28PM -0400, Xiao-Yong Jin wrote:
> I have an idea.  We can leave the original table intact,
> while making a new dynamically generated table derived from
> the original one.

The org-babel idea was the do similar, but again you will not have the
ability to edit the data in the copy.

Often times you want to filter and then edit adjacent items...

Thanks.

------------------------------------------------------------------
Russell Adams                            RLAdams@AdamsInfoServ.com

PGP Key ID:     0x1160DCB3           http://www.adamsinfoserv.com/

Fingerprint:    1723 D8CA 4280 1EC9 557F  66E8 1154 E018 1160 DCB3

Re: Spreadsheet FR

[Xiao-Yong Jin]

On Sun, 4 Apr 2010 15:19:06 -0500, Russell Adams wrote:

> On Sun, Apr 04, 2010 at 12:46:28PM -0400, Xiao-Yong Jin wrote:
>> I have an idea.  We can leave the original table intact,
>> while making a new dynamically generated table derived from
>> the original one.

> The org-babel idea was the do similar, but again you will not have the
> ability to edit the data in the copy.

> Often times you want to filter and then edit adjacent items...

I can imagine some kind of backward propagating of data from
the derived table to the original one by pressing C-u C-c
C-c in the derived table.  I was just thinking that it can
solve the problem of increasing probability of losing data
by miss-typing, because we don't need any extraneous
ellipses, and the original table is always there.  And there
is an advantage of such derived table over an overlay or
Excel that is one can make several different tables from a
single big one according to different criteria.  That might
not introduce more difficulties implementing it than a
direct overlay, but again, I don't know how to implement it.

I guess I am dreaming, now.  But I hope such kind of
functionality is feasible in practice.
-- 
J    c/*    __o/*
X    <\     * (__
Y    */\      <