Issue in 7.6.2 beta

[Morten Leikvoll]

I have this ULP with a redundant bug detector that kicked in all of a
sudden.

Using CONTEXT menu, my test detected if more than one item was in the
ingroup() when executed from context menu from a wire perspective. Of
course this should not be possible.

I tried a few tricks to get rid of the bug, but finally I had to close,
then reopen the project before the bug went away.

Unfortunately I can not recreate it at command, but this may give a hint
what to look for.

This is how the detector looks like:
---------------
found="";
if(schematic) schematic(SCH)//Look for the selected net
{
SCH.sheets(SHT) SHT.nets(NET) NET.segments(SEG) SEG.wires(WIR)
if(ingroup(WIR))
{
if(found!="")//Redundant test
{
dlgMessageBox("Bug detected? Several grouped wires detected in
Sch.MenuContext.Wire context");
exit(0);
}
found=NET.name;
found_page=SHT.number;
}
}
----------------

[Morten Leikvoll]

I just wanted to lift this issue and ask for a confirmation. This
happens all too often to me. If you like me to contribute, please let me
know what to try. Unfortunally this is not a saveable state.

I suspect the group function has some conflicts with the rightclick
context selection that doesnt get resolved properly. Maybe related to
latest SET CONTEXT repair attempts.

[Morten Leikvoll]

I did some deeper digging here and found ingrouped wires inside objects
that were not themself ingrouped, or even at the page where the context
function was triggered.
I was browsing schematic->sheets->nets->segments->wires and segments and
their ancestors were not part of ingroup even if the wire was. I guess
those are false ingrouped.

I see I may add some redundancy to my script to only pick those ingroups
where the parent is also part of the ingroup, but still this is flaky.

[Morten Leikvoll]

Another interesting issue.. Those false ingrouped wires were something I
moved to a different sheet not long ago. When just grouping them again,
the false ingroup flags seems to get cleared.

Maybe the context ingroup function should get a new function name, like
incontext()? I guess its not too late to do that, flaky as it has been.
Logically they are totally different, and using the ingroup() for this
seems to be a horrible shortcut.

[Richard Hammerl]

We tried this and could not reproduce the problem. Do you have any
further news on ths in order to reproduce it reliable meanwhile?


--
Mit freundlichen Gruessen / Best regards
Richard Hammerl
CadSoft Support -- hot...@cadsoft.de
FAQ: http://www.cadsoft.de/training/faq/

[Walter Spermann]

On 09/20/2016 10:15 AM, Morten Leikvoll wrote:

> On 19.09.2016 15:50, Richard Hammerl wrote:
>> Am 15.09.2016 um 12:23 schrieb Morten Leikvoll:
>>> On 15.09.2016 12:16, Morten Leikvoll wrote:
>>>> I did some deeper digging here and found ingrouped wires inside objects
>>>> that were not themself ingrouped, or even at the page where the context
>>>> function was triggered.
>>>> I was browsing schematic->sheets->nets->segments->wires and segments and
>>>> their ancestors were not part of ingroup even if the wire was. I guess
>>>> those are false ingrouped.
>>>>
>>>> I see I may add some redundancy to my script to only pick those ingroups
>>>> where the parent is also part of the ingroup, but still this is flaky.
>>>>
>>>
>>> Another interesting issue.. Those false ingrouped wires were something I
>>> moved to a different sheet not long ago. When just grouping them again,
>>> the false ingroup flags seems to get cleared.
>>>
>>> Maybe the context ingroup function should get a new function name, like
>>> incontext()? I guess its not too late to do that, flaky as it has been.
>>> Logically they are totally different, and using the ingroup() for this
>>> seems to be a horrible shortcut.
>>>
>>
>>
>> We tried this and could not reproduce the problem. Do you have any
>> further news on ths in order to reproduce it reliable meanwhile?
>

> I can reproduce this now. As I thought it seems to be related to moving schematic across pages.
>
> How to reproduce:
> 0-Unzip the files attached using the opensource 7zip tool (free download at 7zip.org)
>
> 1-Use the attached tiny test project
>
> 2-In the schematic editor, run the attached ulp using "run test_wire_context install" to install
> context menu
>
> 3-Group and move everything on page 1 to page 2
>
> 4-Undo
>
> 5-Rightclick the the wire between C3 and C4, get a multiple ingroup item error message and read the
> ingroup() status list in debug.txt.
>
>
We also could reproduce it and fixed it for next beta.
Thanks for your effort!

Regards,
Walter Spermann

--
----------------------------------------------------------------------------------------------------------
Walter Spermann
Software Development
CadSoft Computer GmbH
Pleidolfweg 15
84568 Pleiskirchen
Tel.: 08635/6989-10
www.cadsoft.io
----------------------------------------------------------------------------------------------------------

[Morten Leikvoll]

On 20.09.2016 17:14, Walter Spermann wrote:
> We also could reproduce it and fixed it for next beta.
> Thanks for your effort!

I just found another leftover of this in 7.7.0. I can still not repeat
it, but here is a snapshot of the debug text generated by my script
confirming the same issue:

schematic net SIGNAL1:no
schematic net segment:no
schematic net segment wire:yes
schematic net SIGNAL2:no
schematic net segment:no
schematic net segment wire:yes
schematic net SIGNAL3:no
schematic net segment:no
schematic net segment wire:yes


I think those nets may have been involved in some group mirror function.
That's the best clue I can give now. I will followup as soon as I see
more of it.