[DISCUSS] Handling of ByteEncoding in Parser and Serializer?

[Christofer Dutz <ch...@c-ware.de>]

Hi all,

so in general we have the situation in which we usually have either BigEndian or LittleEndian encoding. Or we have dynamic encoding based on data in the packet itself (PROFINET). Now Ben added something I think we’ll be needing on other drivers too (Modbus) … that the encoding is passed in from the outside.

He solved this by passing in a custom type to the Parser and relying on this being saved in an artificial ParserArgument-Field.

I’m not a big fan of storing parser arguments in fields to be able to access them when serializing, because it sort of adds clutter to the model.

I would love to change things in a way, that if a parser argument of type “ByteEncoding” is defined, that this field is also automatically added to the serializer.

What do you folks think?

Chris

Re: [DISCUSS] Handling of ByteEncoding in Parser and Serializer?

[Christofer Dutz <ch...@c-ware.de>]

Nothing would change here … we would just pass in one context instead of some of the fields we’re currently passing in.
Stuff like the ReadBuffer and the WriteBuffer etc. It’s more syntactic sugar.
Right now, I use ThreadLocals to pass along stuff like the “_curItem”, “_lastItem” etc. I would just move that into a Context type pretty much the same way it’s done in Golang (and how I implemented it in C too)

Chris

From: Łukasz Dywicki <lu...@code-house.org>
Date: Monday, 20. February 2023 at 17:15
To: dev@plc4x.apache.org <de...@plc4x.apache.org>
Subject: Re: [DISCUSS] Handling of ByteEncoding in Parser and Serializer?
Hey Chris,
How does this apply to "with" encoding options Sebastian and you worked
on before?
I guess that in such case, if we speak of modbus and application level
byte encoding, it might be more of runtime setting which could be passed
through connection options and propagated down to generated code. Since
it is resolved at runtime we would not need to allocate additional
fields and rely on connection/runtime context and request byte order
from there.. if possible.

Cheers,
Łukasz

On 18.02.2023 17:48, Christofer Dutz wrote:
> Hi all,
>
> so in general we have the situation in which we usually have either BigEndian or LittleEndian encoding. Or we have dynamic encoding based on data in the packet itself (PROFINET). Now Ben added something I think we’ll be needing on other drivers too (Modbus) … that the encoding is passed in from the outside.
>
> He solved this by passing in a custom type to the Parser and relying on this being saved in an artificial ParserArgument-Field.
>
> I’m not a big fan of storing parser arguments in fields to be able to access them when serializing, because it sort of adds clutter to the model.
>
> I would love to change things in a way, that if a parser argument of type “ByteEncoding” is defined, that this field is also automatically added to the serializer.
>
> What do you folks think?
>
> Chris
>
>

Re: [DISCUSS] Handling of ByteEncoding in Parser and Serializer?

[Łukasz Dywicki <lu...@code-house.org>]

Hey Chris,
How does this apply to "with" encoding options Sebastian and you worked 
on before?
I guess that in such case, if we speak of modbus and application level 
byte encoding, it might be more of runtime setting which could be passed 
through connection options and propagated down to generated code. Since 
it is resolved at runtime we would not need to allocate additional 
fields and rely on connection/runtime context and request byte order 
from there.. if possible.

Cheers,
Łukasz

On 18.02.2023 17:48, Christofer Dutz wrote:
> Hi all,
> 
> so in general we have the situation in which we usually have either BigEndian or LittleEndian encoding. Or we have dynamic encoding based on data in the packet itself (PROFINET). Now Ben added something I think we’ll be needing on other drivers too (Modbus) … that the encoding is passed in from the outside.
> 
> He solved this by passing in a custom type to the Parser and relying on this being saved in an artificial ParserArgument-Field.
> 
> I’m not a big fan of storing parser arguments in fields to be able to access them when serializing, because it sort of adds clutter to the model.
> 
> I would love to change things in a way, that if a parser argument of type “ByteEncoding” is defined, that this field is also automatically added to the serializer.
> 
> What do you folks think?
> 
> Chris
> 
>