gEDA-dev: New autonumber text dialog

[Werner Hoch]

Hi Tomaz,

I hope it's ok if I move some parts of the patch to the list.
https://sourceforge.net/tracker/?func=detail&atid=818428&aid=1589430&group_id=161080

On Thursday 02 November 2006 19:30, Tomaz Solc wrote:
> > Can you put a patch to the sourceforge tracker or somewhere else?
>
> I've uploaded the patch to the sourceforge tracker.
>
> I believe it still needs some work though. I found at least one bug
> in the autonumbering code which I don't know how to fix (see my
> description of the patch).

I've applied the patch and played a little bit:

| * Changed the look of the dialog window. It is now
| closer to Gnome HIG.

I like it, lets hear what others think.

The history of the searchtext is great, but it should not contain dups:
  refdes=R?
  refdes=R?  << previous
  refdes=R?  << previous
  refdes=C?
  refdes=C?

It would be nice if:
  * the searchtext won't be added to the history, if it's the previous 
    element in the history
  * or if the searchtext is already in the history (remove the old one 
    and add the new one at the top of the history)

Going on:
What do you think about exchanging the two scope dropdowns "skip numbers 
found in" and "Autonumber text in". I've at least some "brain" problems 
to use it in that order.

The check button for "overwrite existing numbers" ist much better than 
my ration buttons.

When using the dialog I noticed that the sort order makes the dialog 
really large. Is it possible to put them into a dropdown, together with 
icons?
The sort order is still an option. There would be no need for an extra 
headline then.

| * Separated GUI and backend code a bit and added some
| Doxygen comments. The code in x_autonumber.c is still
| hard to read and even harder to understand. In my
| opinion it should be separated into more smaller functions.

Yes, this way it makes sense to split the functionality.

I know the autonumber code is hard to read. There are lots of options 
and combinations to take care of.

| * Added "remove numbers" functionality.

The numbers are only removed if you use a selection that would renumber 
those elements.
Thus, it only deletes the numbers if you select the "Overwrite existing 
Numbers" option.

I think this option should not walk through the whole autonumber code. 
Just iterate over the {selection, page, hierarchy) and remove the 
matching text elements.
(in the code before Step3 in function autonumber_text_autonumber())

| There is at least one bug in the autonumbering code
| which I am unable to fix: 
| 
| Put three elements on an empty page. Set "refdes=U1",
| "refdes=U2" and "refdes=U?". Now select all of them and
| use the autonumber dialog with the following settings:
| 
| Search for: refdes=U?
| Skip: selected objects
  ^^^^^^^^^^^^^^^^^^^^^^
| Autonumber: selected objects
| Overwrite: no
| 
| Other settings at default values.
| 
| Result: the third element gets numbered U1 (so there
| are two elements named U1 on the page)
| 
| Expected result: the third element should be numbered U3.

In my patch I've used:
  namespace = {hierarchy, page, none}
while you've used:
  skip found numbers = { hierarchy, page, selection }
without changing the autonumbering code. It behaves like my patch but 
you're expecting that it should behave like your dialog. ;-)

The code does not build a list of used numbers if namespace=hierarchy or 
namespace=page --> namespace=none.
see: http://archives.seul.org/geda/dev/Sep-2006/msg00106.html

Ok. I'll play a little bit more on saturday.

Regards
Werner