Ghaith Hachem

> This seems to be a diagnostics severity problem. Invalid assignment should be an error, not a warning - then we also would not enter codegen and no panic would...

The way rust does this is they have 2 types of errors, hard errors that cannot be changed, these are the E ones, and then named errors that can be...

> this is a very good point @tisis2 we should have 2 error lists/categories if a error is norm defined error and codesys error, that cannot be configured! I would...

But how do we show this? Do we make these errors have different codes? And give out an error if the error_config.json ignores one of them? We could introduce a...

Skipping diagnostics is indeed not a good idea, so I'm still open for something like the critical level we had before. But I still think that instead of a panic,...

We need to define the exact behaviour of some of these functions. For example `INT_TO_STRING` would just be an alias for `DINT_TO_STRING` but `LREAL_TO_DWORD` is very strange to me, are...

> IEC61131-3 6.6.2.5.3 see the tables e.g. for LREAL_TO_DWORD its defined that it is a binary transfer of the bits on the right side `LREAL_TO_LWORD` ist defined but `LREAL_TO_DWORD` is...

In addittion, we should use the json pretty_print or to_string_pretty method here so that the json is pormatted without needing to pipe it to a formatter.

So the way we do generic is we look for the types you used in your code, and annotate the method you called with your types and look for it....

> > Hello, after printing the corresponding compiling contexts, I find that `IS_VALID` is annotated to `IS_VALID__REAL`, and `IS_VALID_BCD` is annotated to `IS_VALID_BCD__BOOL`! This is strange for me because I...