Which version of the ST language is supported?

[jubnzv]

Hi,

I'd like to ask you, which ST dialect are you plan to support?

There are many variants of ST language extensions provided by vendors like CoDeSys (ExST), Beckhoff (ExST) and even open-source matiec has it's own extensions for IEC61131-3 2ed. As far as I know, no one uses a pure standard. This is because there are some ambiguities in the standard [1], and some vendors trying to make the language more "user-friendly".

So, I'm wondering, are you aiming to support some vendor's language so that other PLC programs can be recompiled? Or this will be a fresh implementation of pure IEC61131-3 standard?

[ghaith]

Hello,

We are initially interested in supporting the 2nd edition standard features, and incrementally add more features to the compiler as needed (3rd Edition features mainly).
We did not consider explicitly supporting some of the vendor extensions, but this might eventually change as the project evolves.
Our aim for now is to be able to compile most of the Standard ST code, but if such extensions make sense I think we could integrate them

[jubnzv]

Thank you for the reply.

We are initially interested in supporting the 2nd edition standard features, and incrementally add more features to the compiler as needed (3rd Edition features mainly).

That sounds reasonable.

And what is the ultimate goal of the project? Are you planning to create an open source industrial-grade compiler, like matiec?

[ghaith]

We would like to indeed get an industrial-grade compiler. The focus is on compiling ST programs to native code. We are not focusing on the usual configuration that follows, or running these programs in some cyclic manner. An ST program compiled with RuSTy would ideally be linked to a companion program that starts and controls it as needed.
It can however, also be used standalone by creating a FUNCTION main : INT and (for now) passing the compiled .o file to clang/gcc to add the necessary rt0.c calls, or used in unit tests.

[jubnzv]

I got it. Thank you for the clarifications!
It will be interesting to compare the performance of binaries created using LLVM and the CodeSyS compiler. According to a recent paper, this can give a comparable speed.

[riederm]

I agree with ghaith, we're aiming for the standard so far.
The codesys extensions (mentioned here sound very doable. Since codesys is form my perspective the most relevant platform for ST I believe we should definitely set a goal to compile codesys flavored ST-code as well. But I did not investigate all the consequences in detail.

[TheColonel2688]

That link is a bit confusing to be as Expressions are part of 61131-3 3rd Edition. I don't know what is being extended.

The other S= and R= I have never even seen used.

I would be very interested in support for Properties as I use them heavily and there is no way I could port and compile any of my code to RuSTy without them. Also FB_Init method for initialization.