auto-uref is broken

This is probably related:
The auto-uref script does not seem to do anything on copy/paste. To reproduce:

0) activate the auto-uref hook in gschemrc

1) open gschem

2) add a resistor symbol. Its refdes will be set to R2 by the hook

3) select the resistor symbol and copy to buffer with [ctrl-c]

4) paste the buffer with [ctrl-v]+[click-on-canvas]

The refdes of the copied symbol is R2. It should be R3.
(Thanks to Krzysztof for adopting this bug.)

---<)kaimartin(>---

You need to use "copy" command to have refdeses updated on pasted objects.
Gschem does not know if the buffer contents came from a CTRL-C or CTRL-X operation - so for objects in the buffer refdeses are not updated.

I have modified the script to 1) keep track of the refdeses per page and 2) check existing refdeses when page is opened/created.
Changes can reviewed/fetched here: http://repo.or.cz/w/geda-gaf/kokr.git/shortlog/refs/heads/auto-uref

What is the difference between the copy action and [ctrl-c]? Wouldn't gschem be able to replace the number by a questionmark character on [ctrl-c]? A paste would then replace the "?" with the auto generated number.

By the way: I'd welcome a hook that would only do the first of the above steps. That is, unnumber the symbols that were copied (with [ctrl-c]). This would avoid duplicates but still allow for full user control over the new numbers.

---<)kaiamrtin(>---

Right now it's a separate operation. In principle it could work the way you have mentioned, ie. hooks for copy, cut and paste could be provided. But I think this falls into "Wishlist" category ;)

Regarding KMK's second point in comment #4, I use the attached hook for renaming to "U?" etc..

Fix committed to git HEAD: http://git.gpleda.org/?p=gaf.git;a=commit;h=9e3496eb1b753ae9e42dfddd2343892a6618455d

Thanks. No autonumber works as expected on add-component. It does not work on copy. No matter how I copy, the refdes stays the same on paste (yp, [ctrl-c], edit menu). I can't get Peters script to work, either. Feels like the copy-copy-component-hook just does nothing.

This are the gschemrc lines by which I try to activate renumber on copy:

(load-from-path "auto-uref.scm")
(load-from-path "unnumber-refdes.scm")
;(add-hook! copy-component-hook unnumber-refdes)
(add-hook! new-page-hook auto-uref-init-page)
(add-hook! add-component-hook auto-uref)
(add-hook! copy-component-hook auto-uref)

---<)kaimartin(>---

Why can't I edit my comment above? No edit buttons there...
The second sentence was meant to start like this: "Now, autonumber works ..."

---<)kaimartin(>---

On Thursday 26 May 2011 02:14:18 KaiMartin wrote:
> What is the difference between the copy action and [ctrl-c]? Wouldn't
> gschem be able to replace the number by a questionmark character on
> [ctrl-c]? A paste would then replace the "?" with the auto generated
> number.
>

Eww. That breaks quite a lot of the assumptions that users would
normally expect with copy/paste. It would also break user expectations
when pasting to other applications, something that Peter C. and I spent
quite a lot of time thinking about how to support (he has an example
patch somewhere that enables copying from gschem and pasting SVG
graphics into Inkscape). We already support copying from gschem and
pasting into a text editor, by the way.

In general, I'm not totally comfortable with the way that auto-uref
keeps its own table of data which is not guaranteed to match up with the
reality of what's going on in the page. Wouldn't all these problems be
solved if the script just scanned the page each time it needed a new
refdes instead of caching its own often-out-of-date data?

                              Peter

--
Peter Brett <email address hidden>
Remote Sensing Research Group
Surrey Space Centre