[Ktechlab-devel] Zoltan strikes again. =P

[P Zoltan]

On Sat, 17 Oct 2009 08:04:26 +0200, Alan Grimes <agrimes at speakeasy.net>  
wrote:

>
> However, I can't say I'm happy that my comment in ~ElectronicConnector
> was ignored. You are expected to try all kinds of configurations on your
> local machine. However, zoltan comitted code that uncommented code that
> I had commented out for a very good reason, -- it segfaulted!!! =(
>

  It it's not commented out, we have the nasty bug of not really removing  
connectors. It didn't segfalulted there for me (yet?). Instead I have lots  
of invalid reads all over the place ( CircuitICNDocument::joinConnectors,  
ItemDocument::~ItemDocument, ItemDocument::unregisterUID ). I guess the  
document data sructures should be documented, and cleaned up; this  
includes the crash on close in 0.3.7

  Testcase: create a logic in and a logic out, connect them. By changing  
logic in's value, logic out's changes too. Now delete the connector and  
change logic in again. The logic out will change its value, even if it  
shouldn't.

  What should we do about this? Leave the bug in or let it crash? Neighter  
is a good idea...

> The real problem is that QT's spooky signaling system is failing us by
> not communicating the "numPinsChanged" signal to our slot (from
> ECNode::setNumPins). =( Fix that and you should be able to change the
> number of pins on buss-splitters all day long without problems.

  Hmm... there are some issues with resizing connected buses -- like  
ktechlab crashing :(

  Do you have more details about this?

>
> I wish I understood better how this signal and slot business works so I
> could explain in much starker terms why it's so brain-damaged. =P
>

  Well, it's documented:

http://doc.trolltech.com/3.3/signalsandslots.html

  Maybe a "connect" is missing somewhere ?