Cadence-Python Interface

Hello, I'm not familiar with this interface with Matlab but it is currently possible to write a simple communication layer between Cadence and Python using ipcSkillProcess.

The idea is that, since both Python and SKILL allow dynamic evaluation of expressions, one can launch a Python process that sends SKILL commands that in turn produce Python code with data objects. It is not particularly efficient since everything is encoded as text and evaluated at execution time, but it is still fast enough to handle large simulation waveforms or explore schematic hierarchies. It is also possible to have access to the full Cadence API by importing its SKILL functions wrapping them as Python objects and calling them from there, allowing one to create custom Python classes on top as needed. Except for error handling, which can be tricky as you're dealing with two outputs from the two interpreters, putting everything together to support most SKILL data types isn't very time consuming.

Personally I think having this kind of interface at low level, with binary wrappers for these imports rather than going through eval(), would be extremely simple to implement and powerful enough not to care that much about whether a high level Python interface is eventually offered or not (for example you can just run your plotting scripts after simulating without needing Python expressions to be integrated in your ADEXL panel)

Hello, I'm not familiar with this interface with Matlab but it is currently possible to write a simple communication layer between Cadence and Python using ipcSkillProcess.

The idea is that, since both Python and SKILL allow dynamic evaluation of expressions, one can launch a Python process that sends SKILL commands that in turn produce Python code with data objects. It is not particularly efficient since everything is encoded as text and evaluated at execution time, but it is still fast enough to handle large simulation waveforms or explore schematic hierarchies. It is also possible to have access to the full Cadence API by importing its SKILL functions wrapping them as Python objects and calling them from there, allowing one to create custom Python classes on top as needed. Except for error handling, which can be tricky as you're dealing with two outputs from the two interpreters, putting everything together to support most SKILL data types isn't very time consuming.

Personally I think having this kind of interface at low level, with binary wrappers for these imports rather than going through eval(), would be extremely simple to implement and powerful enough not to care that much about whether a high level Python interface is eventually offered or not (for example you can just run your plotting scripts after simulating without needing Python expressions to be integrated in your ADEXL panel)