plc4x-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christofer Dutz <>
Subject Re: API Changes proposal
Date Wed, 05 Sep 2018 20:54:50 GMT
Hi all,

just wanted to give you an update on my progress.

I started with updating the examples and integrations and quickly came to a point, where I
had to continue finishing the API refactoring and the base-driver refactoring.

So I just committed my last changes that should make it possible to build Read & Write-Requests.
I really hope to finish this refactoring in the next two days as it's totally driving me nuts.
For the last few days every dream at night has been dealing with architectural problems, byte
encoding and stuff like that ... that has to change ;-)

Just as an example ... the new S7PlcFieldHandlerTest now runs additional 7178 individual tests
to test all combinations of Java and S7 type combinations and their respective value ranges
(MIN, MAX, 0, Some random value).
I still need to implement the temporal types Time, Date and DateTime, and test the "ULWORD"
types, but I guess most should be somewhat usable. 
Wouldn't have thought that the Write direction was so much more work than the Read path.

So much for the Update ...


´╗┐Am 03.09.18, 13:53 schrieb "Julian Feinauer" <>:

    Hi Chris,
    exactly, that was my point (sorry for writing it not out clearly).
    We can do it that way the only thing we are "loosing" is to know whether bit was given
by the user explicitly or not.
    I dont know if we need this after parsing is finished anymore.
    So we can also do it your way.
    Am 03.09.18, 10:47 schrieb "Christofer Dutz" <>:
        Hi Julian,
        had to think a little to get your point ... but think I have ... In my example it's
a primitive "short" and in yours it's "Short" as nullable non-primitive-type.
        I don't technically prefer any of the two options. Physically the bit-offset of any
non-bit type is always 0 and not null (as in undefined) as every non-bit value always starts
at bit 0 ... so for that reason I would prefer "short" with default 0 over the "Short" nullable
version. And this way we don't have to add null-checks in the code. But as I said ... that's
a very slight preference for that option.
        Am 03.09.18, 10:27 schrieb "Julian Feinauer" <>:
            Hi chris,
            I agree with your S7 field except for one little change.
            How do we proceed with the (optional) bit offset?
            I made it "Short" with the contract that null indicates no offset given.
            Another alternative would be to make it 0 as default or even Optional.
            I would prefer to have it nullable, what do you think?
            With the rest I'm fine but as this is part of our internal API now I think we
also have more freedom with evolving them as its not visible to users.
            For all other parts of your proposal +1 from me.
            Am 30.08.18, 10:15 schrieb "Christofer Dutz" <>:
                Hi all,
                especially @Julian ... could you please have a look at that I did with the
S7Field [1]?
                Also there is a unit-test that should allow adding more statements and testing
everything is working ok [2].
                Does this match your idea on [3]? Looking at your addresses, I think that
I might have not quite got it ... is there always a "D" as first part after the "."? I always
read it as "DB" like Data Block ... but seeing DX and SW makes me wonder ... a quick check
in my TIA shows me the address of a Boolean field in a Data Block is "%DB1.DBX38.1" ... which
one is correct?
                As we're no longer constructing the objects themselves in the API, I took
the liberty to simplify the field objects so we now only have one type for S7.
                Am 28.08.18, 12:23 schrieb "Christofer Dutz" <>:
                    Hi all,
                    I just pushed changes to my API refactoring branch ... so far I only adjusted
the API module and added an example using the changed API.
                    To have a look, please go to [1] ...
                    General changes I implemented while working on the refactoring itself.
I did notice, that my current proposal "chris-2" did 
                    Having to inject the type conversion code would have made it necessary
to inject a converter which didn't feel right. So I changed the API to be purely interface
                    In order to be able to construct these objects I also added builders for
                    I asked a few people here what they think, and most liked the simplicity
and didn't have any WTF experiences (Which seems to be a good thing as I did have to explain
a lot with the current API)
                    Quick Feedback highly appreciated as I will start implementing DefaultPlcReadRequest
& Co (in driver-base ... together with the builders) after that I'll start migrating the
                    Right now having a look a named example [1] would be a good start ...

                    Second would be a deeper look into the API module [2].
                    Would be a shame to waste that time and effort if you think the changes
suck (or are less than optimal as non-Germans would probably call them ;-) ) .
                    Am 27.08.18, 09:57 schrieb "Christofer Dutz" <>:
                        Ups ... after reloading .. I just saw Julians Proposal pop up ...
haven't looked into that ...
                        Will do that right away.
                        Am 25.08.18, 15:52 schrieb "Christofer Dutz" <>:
                            Hi Julian,
                            version 2 should now be quite different ... I started reworking
my original proposal and decided to revert that an start a second proposal.
                            My first did address some parts needing cleaning up, but I still
wasn't quite satisfied with it. So I did another more radical refactoring.
                            If you reload the second there should be a lot of differences.
                            I just hit "save" a few minutes ago however ... but now I'm quite
happy with it. So please have another look at the second proposal. 
                            And please, maybe add your own proposal ... my versions are just
Brainstorming from my side.
                            My favorite is currently "Chris' Proposal 2" ;-)

View raw message