PCBnew/EESchema: Net naming convention breaks older designs.
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
KiCad |
Fix Committed
|
Medium
|
Jon Evans |
Bug Description
We're migrating our pcb design from kicad 5.1 to the current nightly. However, it looks like the newer version of kicad resolves netnames differently than 5.1.
We have a very large hierarchical design (>110 sheets, 5k nets).
When using the hierarchical sheets, and importing the new design, kicad seems to be changing how it determines netnames from hierarchies.
This is causing the pads to have a new netname that is different than the original netname for the various traces. This is especially true for our modular design.
This makes it really hard to do layout because then I would have to reroute or reassign all the nets that have changed, and this is happening across all our DDR4 ram blocks!!
Is there a way to make the traces infer their netnames based on the connected lands? Or else have the netname inferred in the same way older versions did?
tags: | added: eeschema |
Changed in kicad: | |
status: | Incomplete → Confirmed |
Changed in kicad: | |
status: | Confirmed → In Progress |
Hi Madeliene,
There have been some changes made to the way net names are determined in order to make the system more predictable and enable new features. However, this change is fairly new and it is possible there are still bugs in it.
Can you please share details about what changes occurred, and what the desired behavior is for you? I will check to see if an improvement is possible. If you are able to share your design privately with me it will help speed up fixing this. If not, some more detailed description of an example change will help.
Also, to be clear, are you saying that the net names have changed (but the connection between pads is still the same) or that the actual connections have also changed? In other words, when you import the netlist from schematic into PcbNew, do connections break / new ratsnest lines appear? If that's the case, this is a different problem.
Thanks