plc4x-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ben Hutcheson <ben.hut...@gmail.com>
Subject Re: [MODBUS][DISCUSS] Improving handling of datatypes ...
Date Fri, 11 Sep 2020 21:04:41 GMT
Hi,

After creating a PR for the modbus data types and moving the max/min checks
and data conversion tasks to the individual PLCValue classes, the
ModbusFieldHandler class has been reduced to a bunch of cases which map a
Java Datatype to a PLC Datatype. Looking at the S7FieldHandler class it
seems to be doing the same.

What are your thoughts on creating a new section within the mspec file to
allow mapping Java datatypes to PlcValue types and then automatically
creating the FieldHander class for each protocol? The mspec section would
list all the Fields available for the protocol as well as datatypes e.g.
Modbus would have the coil/discrete/input/holding/extended fields and map
the Java Integer to INT,UINT,DINT,REAL, map Floats to REAL types, etc..

This would allow protocols to override mappings to PLCValue classes when
needed as well.

Kind Regards

Ben

On Tue, Sep 8, 2020 at 7:52 PM Ben Hutcheson <ben.hutche@gmail.com> wrote:

> Hi Chris,
>
> I'm not sure I follow your last email,
>
> I have pushed a commit to *hutcheb:Feature/ModbusDataType
> <https://github.com/hutcheb/plc4x/tree/Feature/ModbusDataType>, *which
> includes some of the changes I've proposed.
>
> 1) Within
> build-utils/language-java/src/main/resources/templates/java/data-io-template.ftlh,
> I've added a cast to the PLCValue so that we can use methods specific to
> the PLCValue types to decode input from th PLC not just the methods defined
> in the interface.
> 2) I've added the initial framework for adding Modbus datatypes within the
> address string. This is able to support the INT (r/w) and UINT (w)
> datatypes when parsing a Java String (I have to clean it up and fix up
> handling of PLCList types)
> 3) I've updated the Modbus.mspec to include the INT datatype within the
> dataio section, I would have to add the rest of the datatypes in the same
> section.
> 4) I've added PlcINT and PlcUINT classes and have moved the handling of
> parsing different Java Datatypes into these, as well as some handling of
> decoding within the PlcINT classes. Along with the last point this allows
> us to decode the INT datatype within the DataItemIO.java file.
>
> Let me know what you think?
>
> Kind Regards
>
> Ben
>
> On Tue, Sep 8, 2020 at 11:44 AM Christofer Dutz <christofer.dutz@c-ware.de>
> wrote:
>
>> Hi Ben,
>>
>> well the PlcValue types are based on the types used in the language they
>> are used in. In PLC4C, which supports signed and unsigned types, I
>> implemented them differently.
>> Initially I thought that if the PlcValues would simply have a byte-array
>> and a type, then it would be universal ... but I thought accessing the
>> values would require more cpu and memory.
>>
>> But given the fact that if we had such generic PlcValues, we could use
>> them in all drivers ... sort of makes them more interesting.
>>
>> We could have multiple implementations of them ... so we could have these
>> DefaultPlcValue implementation for the normal 61131 datatypes but give
>> drivers the opportunity to provide customizations for special types.
>>
>> What do you think?
>>
>> Chris
>>
>>
>>
>> Am 08.09.20, 14:44 schrieb "Ben Hutcheson" <ben.hutche@gmail.com>:
>>
>>     Hi Chris,
>>
>>     The case where PLC data types are mapped to 16 bit registers isn't a
>>     special case, as the Modbus protocol only supports 16 bit registers
>> but it
>>     doesn't specify what format the data should be in,any PLC data type
>> can be
>>     mapped to these and then passed over Modbus. It would definitely be
>> handy
>>     to be able to convert any of the PLC datatypes back from Modbus 16-bit
>>     registers to whatever datatype they should be. I've done some work
>> already
>>     on the specific case of mapping a Java String to the
>> PLCInteger/PLCShort
>>     classes and can expand it basing it off the S7 functions that do the
>> same.
>>     We can then look at the differences and consolidate them if possible?
>>
>>     However it would be great to be able to add unambiguous classes to
>>     plc4j/api/src/main/java/org/apache/plc4x/java/api/value for each of
>> the IEC
>>     datatypes instead of these being based off Java types.
>>
>>     Using the same format as the S7 protocol the address string would be
>>     holding-register:1000:REAL[10] instead though, I don't think this will
>>     cause issues if we make it optional and default to INT or BOOL as
>> required.
>>
>>     Kind Regards
>>
>>     Ben
>>
>>
>>
>>
>>
>>
>>
>>     On Tue, Sep 8, 2020 at 8:10 AM Christofer Dutz <
>> christofer.dutz@c-ware.de>
>>     wrote:
>>
>>     > Hi Ben ...
>>     >
>>     > Sorry for my email... I should really finish reading before
>> replying :-(
>>     >
>>     > About your suggestion to add it to the type to the ModbusField ... I
>>     > agree. But I would simply extend the Address string the same way
>> the TIA
>>     > addresses were extended.
>>     > Cause the problem I'm seeing is that for the 61131 datatypes we can
>>     > provide an enum, but not for any custom types the driver might be
>>     > supporting. With a String we can keep it simple ... check if the
>> type is a
>>     > 61131 type -> if yes: do the default handling. If it's not, use
>> some driver
>>     > specific encoding/decoding.
>>     >
>>     > In my particular case the Float values were also 4 byte values ... I
>>     > manually solved it by requesting 2 element arrays for every float
>> address
>>     > (The map of reigsters only used only even register numbers, so that
>>     > worked). Then I manually converted the 2 x 2bytes into a 4byte
>> float with
>>     > this code:
>>     >
>>     >             // Decode the first 10 shorts as 32bit floats ...
>>     >             StringBuilder sb = new StringBuilder();
>>     >             for(int i = 0; i < 10; i++) {
>>     >                 int short1 = value.getIndex(i * 2).getInteger();
>>     >                 int short2 = value.getIndex((i * 2) +
>> 1).getInteger();
>>     >                 WriteBuffer wb = new WriteBuffer(4);
>>     >                 wb.writeInt(16, short1);
>>     >                 wb.writeInt(16, short2);
>>     >                 final byte[] data = wb.getData();
>>     >                 int intVal = (data[3] & 0xFF) | ((data[2] & 0xFF)
>> << 8) |
>>     > ((data[1] & 0xFF) << 16) | ((data[0] & 0xFF) << 24);
>>     >                 final float v1 = Float.intBitsToFloat(intVal);
>>     >                 sb.append(v1).append(", ");
>>     >             }
>>     >
>>     > Ok in this case I read 10 floats by reading an array of 20
>> registers.
>>     > Guess it should be even simpler with 32bit integers
>>     >
>>     > But I guess it does the job ... not pretty, but it works and I would
>>     > definitely love to have something similar in the drivers to handle
>> this if
>>     > I said "holding-register:1000[10]:REAL"
>>     >
>>     > Chris
>>     >
>>     >
>>     > Am 08.09.20, 13:59 schrieb "Christofer Dutz" <
>> christofer.dutz@c-ware.de>:
>>     >
>>     >     Hi Ben,
>>     >
>>     >     if you have followed the latest discussions here ... currently
>> the
>>     > modbus protocol supports 16bit integer values for registers and
>> Boolean
>>     > values for coils.
>>     >     The protocol doesn't directly support anything else. However I
>> have
>>     > seen that vendors use arrays of registers to support higher level
>> types. In
>>     > my special case the PLC vendor used two 16bit integers to provide
>> 32bit
>>     > float values.
>>     >
>>     >     We will probably start adding support for the basic IEC 61131
>>     > datatypes (https://en.wikipedia.org/wiki/IEC_61131-3) but I think
>> this
>>     > will only apply for the Bool, Bit-String, Signed and Unsigned
>> Integers as
>>     > well as Real values. Durations, Times, Dates and especially Strings
>> will be
>>     > problematic as it seems every vendor uses a different encoding of
>> these.
>>     >
>>     >     Chris
>>     >
>>     >
>>     >     Am 08.09.20, 13:52 schrieb "Ben Hutcheson" <
>> ben.hutche@gmail.com>:
>>     >
>>     >         Hi,
>>     >
>>     >         Looking at the example I had with the hello-world-write
>> example
>>     > trying to
>>     >         write a String to the Modbus protocol.
>>     >
>>     >         ava -jar
>> plc4j-hello-world-plc4x-write-0.8.0-SNAPSHOT-uber-jar.jar
>>     >         --connection-string modbus://127.0.0.1:5467
>> --field-addresses
>>     > 40010
>>     >         -field-values 32000
>>     >
>>     >         The value 32000 then gets cast as a String and is passed to
>> the
>>     >         encodeString function in ModbusFieldHandler. (It doesn't
>> have one
>>     > it gets
>>     >         passed to the DefaultFieldHandler). To get this to work
>> properly
>>     > we would
>>     >         need to define a datatype to cast the String to, S7 does
>> this
>>     > already and
>>     >         with the consensus that IEC 61131 data types be used where
>>     > possible then I
>>     >         see no reason we shouldn't add the datatype field to the
>>     > ModbusField. If we
>>     >         are doing this for these protocols, then why don't we add
>> the
>>     > datatype
>>     >         field to the PLCField interface, this would allow us to use
>> the
>>     > same
>>     >         default encodeString (We may have to map the protocol
>> specific
>>     > datatypes to
>>     >         IEC 61131 here, the same internalEncodeString might be
>> better)
>>     > functions
>>     >         for S7 and Modbus by casting the input to a PLCField
>> instead of a
>>     > S7Field
>>     >         or ModbusField.
>>     >
>>     >         When encodeString is then called we can use the datatype
>> that is
>>     > passed to
>>     >         it to encode it to a PLCInteger/PLCFloat/etc.., the same
>> way the S7
>>     >         protocol does it. S7 however only allows Strings to be
>> passed as
>>     > string
>>     >         related variables (CHAR, WCHAR, STRING and WSTRING), I
>> don't see
>>     > any reason
>>     >         we can't expand this to be able to pass any PLC datatype
>> from a
>>     > string.
>>     >
>>     >         The next problem I had was to then be able to parse the
>> PLCInteger
>>     > class to
>>     >         a byte array within the fromPLCValue function. For the UINT
>>     > datatype for
>>     >         the Modbus protocol, this is a 2 byte array however the
>> PLCInteger
>>     > class
>>     >         contains it as a 4 byte array. in the fromPLCValue function
>> there
>>     > doesn't
>>     >         seem to be a way to test to see if it is an UINT or a 4 byte
>>     > integer such
>>     >         as a DINT. If we were to change the classes in
>>     >         plc4j/api/src/main/java/org/apache/plc4x/java/api/value to
>> match
>>     > the IEC
>>     >         61131 datatypes then we should be able to parse them
>> correctly in
>>     > Modbus
>>     >         fromPLCValue function.
>>     >
>>     >         I have probably missed a few things out, like how the
>>     > TransportSize class
>>     >         for the S7 protocol works, and we may have to settle for
>> mapping
>>     > protocol
>>     >         specific types to IEC61131 types in the encode functions in
>> each
>>     > protocol,
>>     >         what are your thoughts?
>>     >
>>     >         Kind Regards
>>     >
>>     >         Ben
>>     >
>>     >
>>     >
>>     >
>>     >         On Sun, Sep 6, 2020 at 11:25 AM Christofer Dutz <
>>     > christofer.dutz@c-ware.de>
>>     >         wrote:
>>     >
>>     >         > Hi all,
>>     >         >
>>     >         > well I doubt that we could really centrally handle this.
>>     >         > If you have an idea how to unify this, I'd be happy for a
>>     > suggestion.
>>     >         > I guess most will have to be handled in the individual
>> drivers
>>     > themselves.
>>     >         >
>>     >         > But I really would like all drivers to support the same
>> base-set
>>     > of
>>     >         > datatypes.
>>     >         >
>>     >         > Chris
>>     >         >
>>     >         >
>>     >         >
>>     >         > Am 06.09.20, 17:02 schrieb "Cesar Garcia" <
>>     > cesar.garcia@ceos.com.ve>:
>>     >         >
>>     >         >     Hello,
>>     >         >
>>     >         >     I think it can be applied from several points of
>> view, the
>>     > one
>>     >         > typified as
>>     >         >     you indicate (HREGISTER: REAL) or (HREGISTER [2]),
>> each one
>>     > has its
>>     >         >     advantage.
>>     >         >
>>     >         >     To be able to handle the types indifferently
>> (specifically
>>     > between
>>     >         > Modbus
>>     >         >     and S7), I modify the field of the Modbus driver so
>> that it
>>     > strictly
>>     >         > adapts
>>     >         >     to the PLC4X API. This prevents the user's App from
>> having
>>     > to implement
>>     >         >     code to interpret the content of the records.
>>     >         >
>>     >         >     This is implemented in version 0.6.1 of  PLC4X, but
>> the new
>>     > version is
>>     >         >     something else.
>>     >         >
>>     >         >     Best regards,
>>     >         >
>>     >         >     El dom., 6 sept. 2020 a las 10:03, Ben Hutcheson (<
>>     >         > ben.hutche@gmail.com>)
>>     >         >     escribió:
>>     >         >
>>     >         >     > Hi,
>>     >         >     >
>>     >         >     > I ran into this issue this morning when using the
>>     > hello-world-write
>>     >         > example
>>     >         >     > and trying to write to a Modbus connection. It
>> looks like
>>     > it assumed
>>     >         > the
>>     >         >     > input value is a string whereas the Modbus protocol
>>     > doesn't have
>>     >         > support
>>     >         >     > for it yet.
>>     >         >     >
>>     >         >     > I was thinking about expanding
>>     >         >     >
>>     >         >     >
>>     >         >
>>     >
>> github/plc4x/plc4j/spi/src/main/java/org/apache/plc4x/java/spi/connection/DefaultPlcFieldHandler.java
>>     >         >     > to include default handlers for the various IEEE
>> 61131
>>     > data types,
>>     >         > the
>>     >         >     > specific protocols can then override them as
>> necessary?
>>     >         >     >
>>     >         >     > Kind Regards
>>     >         >     >
>>     >         >     > Ben
>>     >         >     >
>>     >         >     > On Wed, Sep 2, 2020 at 9:36 AM Christofer Dutz <
>>     >         > christofer.dutz@c-ware.de>
>>     >         >     > wrote:
>>     >         >     >
>>     >         >     > > Hi Julian,
>>     >         >     > >
>>     >         >     > > I agree ... if one driver would define an "INT"
>> as 32bit
>>     > integer
>>     >         > and the
>>     >         >     > > others would treat it as 16bit ... that could
be a
>>     > problem.
>>     >         >     > > Perhaps having a statement of the project that
we
>> use
>>     > the IEC
>>     >         > 61131 types
>>     >         >     > > as a basis and if you want to use a given
>> protocols
>>     > different
>>     >         > types, that
>>     >         >     > > you can prefix them ..
>>     >         >     > >
>>     >         >     > > Assuming a driver for the famous "HURZ" protocol
>> would
>>     > use 32bit
>>     >         > INTs,
>>     >         >     > > then an "INT" could reference the 16bit version
>> and a
>>     > "HURZ_INT"
>>     >         > could be
>>     >         >     > > the 32bit version?
>>     >         >     > >
>>     >         >     > > Chris
>>     >         >     > >
>>     >         >     > >
>>     >         >     > >
>>     >         >     > >
>>     >         >     > > Am 02.09.20, 15:26 schrieb "Julian Feinauer"
<
>>     >         >     > > j.feinauer@pragmaticminds.de>:
>>     >         >     > >
>>     >         >     > >     Hi,
>>     >         >     > >
>>     >         >     > >     agree with your suggestion!
>>     >         >     > >     Although we have to be careful to not mix
it
>> up with
>>     > specific
>>     >         >     > > implementations of datatypes in some drivers.
>>     >         >     > >
>>     >         >     > >     Julian
>>     >         >     > >
>>     >         >     > >     Am 02.09.20, 15:21 schrieb "Christofer Dutz"
<
>>     >         >     > > christofer.dutz@c-ware.de>:
>>     >         >     > >
>>     >         >     > >         Hi all,
>>     >         >     > >
>>     >         >     > >         today I was at a customer’s site and
used
>> the
>>     > Modbus
>>     >         > driver to
>>     >         >     > get
>>     >         >     > > data. This generally worked fine.
>>     >         >     > >         The thing however I found a little
>> complicated,
>>     > was that
>>     >         > the PLC
>>     >         >     > > seemed to offer all values as 32Bit Floating
point
>>     > values.
>>     >         >     > >         So in order to correctly read them, I
had
>> to
>>     > read an array
>>     >         > of two
>>     >         >     > > consecutive shorts and then manually convert
them
>> into a
>>     > float.
>>     >         >     > >
>>     >         >     > >         I bet we can do this better.
>>     >         >     > >
>>     >         >     > >         So I thought … how about we use the
same
>>     >         >     > > https://en.wikipedia.org/wiki/IEC_61131-3
>> datatypes we
>>     > are
>>     >         > already using
>>     >         >     > > in other drivers and for example if you write:
>>     >         >     > >
>>     >         >     > >         holding-register:1:REAL
>>     >         >     > >
>>     >         >     > >         it would automatically use modbus to
read
>> an
>>     > array of two
>>     >         > shorts
>>     >         >     > > and to internally convert these to one REAL/float.
>>     >         >     > >
>>     >         >     > >         What do you think? I think we could
>> probably do
>>     > this in
>>     >         > most
>>     >         >     > > drivers.
>>     >         >     > >
>>     >         >     > >         Chris
>>     >         >     > >
>>     >         >     > >
>>     >         >     > >
>>     >         >     > >
>>     >         >     > >
>>     >         >     >
>>     >         >
>>     >         >
>>     >         >     --
>>     >         >     *CEOS Automatización, C.A.*
>>     >         >     *GALPON SERVICIO INDUSTRIALES Y NAVALES FA, C.A.,*
>>     >         >     *PISO 1, OFICINA 2, AV. RAUL LEONI, SECTOR
>> GUAMACHITO,*
>>     >         >
>>     >         >     *FRENTE A LA ASOCIACION DE GANADEROS,BARCELONA,EDO.
>>     > ANZOATEGUI*
>>     >         >     *Ing. César García*
>>     >         >
>>     >         >     *Cel: +58 414-760.98.95*
>>     >         >
>>     >         >     *Hotline Técnica SIEMENS: 0800 1005080*
>>     >         >
>>     >         >     *Email: support.aan.automation@siemens.com
>>     >         >     <support.aan.automation@siemens.com>*
>>     >         >
>>     >         >
>>     >
>>     >
>>     >
>>
>>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message