I am a junior engineer at KGM working in the automotive domain, utilizing Vector CANoe and CAPL (Communication Access Programming Language) for real-time diagnostics, simulation, and test automation.
I would like to ask for help in coding CAPL and configuring the interface in Simulink for automated verification using ChatGPT. But currently, ChatGPT shows major limitations when supporting CAPL-based development. These include:
- Insufficient understanding of CAPL-specific syntax and semantics
- Lacks support for
on envVar,testcase, andsystem variableconstructs - Tends to default to C-like syntax which is incompatible with CANoe compilation
- In most cases, ChatGPT exhibits symptoms where it thinks it has written correct code for CAPL after coding based on C language-based data.
- Poor handling of system/environment variables
- Functions like
WriteEnvVar(),ReadSysVar()are not used properly or type-safely - Does not recognize race conditions or value timing issues in the CAPL runtime model
- Inability to detect compile-time errors in CANoe
- ChatGPT often generates code that compiles incorrectly in CANoe but claims it will run successfully
- No support for static validation (race conditions, code metrics, runtime errors)
- Cannot detect context-specific issues such as symbolic references or runtime context mismatches
Request:
Please improve CAPL support by:
- Learning Vector CANoe’s execution and compilation environment
- Supporting error-aware CAPL code generation
- Maintaining consistent user-defined CAPL coding guidelines across sessions
- Enabling detection of syntax errors that would cause CANoe compilation to fail
CAPL is critical for many OEMs and Tier-1 suppliers worldwide. Accurate support would greatly enhance ChatGPT’s utility in the automotive engineering domain.
I hope ChatGPT is more friendly to industrial programming than other AI models. Thanks.