AFAIR we have only one mspec which is making use of curPos and its mqtt.
S7 does only in one place. Long time ago I have done some experiments
with m-bus, yet they are stale for more than 2 years.
The main question I do raise - do we need a curPos at all? It may not be
portable across all languages. Having curPos and curBitPos double
effort. Maybe instead of promoting these to be special variables we can
generalize these to a parser/serializer context object which could be
referenced from mspec and managed in conjunction with either read or
write buffer?
Cheers,
Łukasz
On 9/17/25 20:09, Sebastian Rühl wrote:
no the question was, does this then turn out to be 4 or 5?
(also something that jumped to my mind if we could overflow as instead of max
int we only support max int / 8 as max :D)
- Sebastian
On 2025/09/17 15:28:51 Christofer Dutz wrote:
Well … currently there is no case where this would not turn out to be an
integer as all places are 8-bit aligned.
Theoretically even an error would be good, because what the hell would you do
with getting „4.5433“?
Chris
Von: Sebastian Rühl <[email protected]>
Datum: Mittwoch, 17. September 2025 um 16:19
An: [email protected] <[email protected]>
Betreff: Re: AW: AW: [DISCUSS] Make "curPos" a bit-position?
sounds good to me. Just quick sanity check: does this always ceil?
- Sebastian
On 2025/09/17 14:09:07 Christofer Dutz wrote:
I would actually opt for updating any expressions from „curPos“ to
„(curPosInBits/8)"
Chris
Von: Sebastian Rühl <[email protected]>
Datum: Mittwoch, 17. September 2025 um 15:59
An: [email protected] <[email protected]>
Betreff: Re: AW: [DISCUSS] Make "curPos" a bit-position?
ack. Do you think we should name the other method then curPosInBytes too?
- Sebastian
On 2025/09/17 13:54:42 Christofer Dutz wrote:
I see it the same way … but I think we can make things simpler, if we name
things differently.
So „curPosInBits“ would be a better way of naming it, as it explicitly states
what we’re counting.
In most of IT it’s always bytes.
Chris
Von: Sebastian Rühl <[email protected]>
Datum: Mittwoch, 17. September 2025 um 15:50
An: [email protected] <[email protected]>
Betreff: Re: [DISCUSS] Make "curPos" a bit-position?
I would even go that far and always use bit positions and only use byte
positions for box rendering and as utility method. Also making the type more
explicit in the name is a good idea I think.
- Sebastian
On 2025/09/16 19:40:32 Christofer Dutz wrote:
Hi all,
Just came back from Minneapolis this morning and was continuing my work on
cleaning up PLC4X. Here I noticed one thing, that could cause problems.
It seems that „curPos“ in mspec refers to the byte-position in the byte-buffer.
This has never caused issues before, but I know there are protocols out there,
that have non byte aligned complex objects. I think in firmata there were types
that were only 4 bits long.
By sticking to the bytePosition, we will have issues modeling such protocols
fully.
If however, we updated our mspecs to expect curPos to be a bit-position, we
would avoid this type of problem.
What do you think? Should we update this? I also think that it might be
beneficial to be a bit more explicit. We could add a „curBitPos“ and rename the
existing „curBytePos“ for compatability reasons.
Chris
