Hi Chris, we have pushed all local changes in Git and will not make any changes in the next days.
I think we now have the build process stable for C++. At the end you have to decide if we want to convert the branch back to devolop. I have no problem with that anymore. Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Freitag, 12. Juli 2019 15:54 An: [email protected] Betreff: Re: [C++] Status Update? Hi Markus, thanks for that :-) Highly appreciated :-) Also thanks for testing it on all the platforms. I know this is sometimes quite a hassle, but it simplifies things for others. Perhaps we should think of changing our build to spawn the building also on a Windows and a Mac node in parallel to the Linux node we also do the testing on. Are there any objections from me merging the current state back to develop? I don't have to delete the branch you're working on ... I just don't want the branches to diverge too much. In general I would also like to start some experiments with the code generation and with a thrift code generation. Chris Am 12.07.19, 14:50 schrieb "Markus Sommer" <[email protected]>: Hi Chris, I have now implemented the PLatfrom-specific part for Linux as well and compiled the sources for Linux and Windows. Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Freitag, 12. Juli 2019 09:05 An: [email protected] Betreff: Re: [C++] Status Update? Hi Markus, I just noticed that I accidentally committed the changes where I commented out those parts, that don't build on Mac and Linux ... so please have a look at commit: 1452678d5ee9dd68c63e035c0963078e0a3bfe71 Sorry for that ... but at least this way the build works for all platforms. How about merging back the changes of feature/s7-cpp and leaving it active as I would like to have the changes to the build I did in that branch in develop too? As plc4cpp is just as unusable as plc4py and plc4net I don't think there should be any problems. What do you and the others think? Chris Am 11.07.19, 11:53 schrieb "Christofer Dutz" <[email protected]>: Hi all, so I noticed that in the PLC4X build the Thrift compiler stopped working. The strange thing, that was driving me nuts was that nothing had changed from a configuration point of view. In the end I managed to find out that my efforts to building Boost had the boost 1.70 version installed on my system (I must have built once without specifying a prefix). I found some comments on Thrift having problems with 1.70 and a comment in the repo for the unreleased version of Thrift, where they simply disable Boost if 1.70 is found. I managed to replicate this behavior and now Thrift seems to be working again: Yay! ... only took me about 24 hours to figure out (including a sleepless night). But here I also had an Idea how we can speed up the build with not having to build Boost every time: - I created a new module "boost" in "tools" which builds boost and packages everything into a zip file. - I unpack that maven dependency when building plc4cpp (That's a loooooooot faster than compiling) - I added a profile "with-boost" where you can enable building of the library so if you don't enable it, it will take/or download the corresponding version from Maven. Will play around with this option ... keep your fingers crossed. Chris Am 10.07.19, 15:57 schrieb "Christofer Dutz" <[email protected]>: Hi Markus, well currently I think we should use gcc as Visual Studio isn't available on all platforms. And I'm currently configuring the maven build 64 bit only ... but I guess we can adjust that as soon as we need 32 bit. Also I'd like to report, that I needed to do some changes (Add some "-fPIC" flags to the boost compilation as well as to the build itself) or it wouldn't build on Linux. I double checked that these changes didn't break anything. So now the build passes on Linux, Mac and Windows (I commented out the Windows-Specific code parts for that) Chris Am 10.07.19, 15:47 schrieb "Markus Sommer" <[email protected]>: Hi Chris, great. From my point of view, we still have two points to do when it comes to generation: Selection of the compiler (gcc or Visual studio) Selection of architecture x86 or x64 But they can also go on later. Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Mittwoch, 10. Juli 2019 11:30 An: [email protected] Betreff: Re: [C++] Status Update? Hi Markus, so it seems now, with the newer CMake it appears to work correctly with setting in the CMakeLists.txt file, so I removed the Environment variable setting in the pom and moved that to the CMake files. On my Mac's CLion it worked perfectly, even if I can't seem to set breakpoints ... currently trying for Linux ... Chris Am 10.07.19, 08:38 schrieb "Christofer Dutz" <[email protected]>: Hi Markus, I manually set the variables in clion and was able to build. But as the path now no longer contains any version information, we should be fine. When I first tried it, cmake wouldn't accept the values of I specified them in the CMakeLists.txt file, but I explicitly read it in the documentation so it might have been a bug and works now. I'll try that out as soon as I'm in the office. And I agree, it helps setting things up a lot. Chris Holen Sie sichOutlook für Android<https://aka.ms/ghei36> ________________________________ From: Markus Sommer <[email protected]> Sent: Wednesday, July 10, 2019 8:24:17 AM To: [email protected] Subject: AW: [C++] Status Update? Hi Chris, creating with Maven works very well. I noticed the following points: 1. How can I select the compilers (Visual C++ or gcc) for Maven? 2. The variables BOOST_INCLUDEDIR and BOOST_LIBRARYDIR are passed by maven. If plc4cpp is compiled in a development environment like CLion, the value of the variables is missing and the driver cannot be created. Should we clean up a query in CMake and set it locally? Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de<http://www.isb-fn.de> Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Dienstag, 9. Juli 2019 21:41 An: [email protected] Betreff: Re: [C++] Status Update? Yup ... my latest commit fixed the problem. Guess you can concentrate on making the code compile for linux and mac ( Chris Am 09.07.19, 21:32 schrieb "Christofer Dutz" <[email protected]>: Think I fixed it ... was missing a linker assignment in the CMakeList.txt Chris Am 09.07.19, 17:27 schrieb "Markus Sommer" <[email protected]>: Hello, Chris, Thank you so much for your great work. I'm checking out the LinkerError tonight. Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de<http://www.isb-fn.de> Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Dienstag, 9. Juli 2019 16:59 An: [email protected] Betreff: Re: [C++] Status Update? Hi Markus, Ok so I fixed another few things that weren't working correctly. It wasn't correctly finding the boost libraries as it wan't setting the arch to x64 correctly. Now it's finding the libraries and manages to compile everything and also to almost finish linking. Now it builds to 97% and then fails when linking the s7.dll ... complaining about undefined references to mainly the constructors and destructors of ExLog and finally dies because of an "bad reloc address 0x0 in `.data'" Now I'm really starting to feel lost ... hopefully you'll manage to sort this out ... otherwise I'll try continuing tomorrow. Chris Am 09.07.19, 15:11 schrieb "Christofer Dutz" <[email protected]>: No need to look, just update ... I had a look and noticed the declspec needed two underscores ... as soon as I fixed that compilation passes on my windows VM. However the linking seems to fail. As far as I got it from my google-foo ... it seem that the winsock library needs to be linked in differently for gcc as wich the visual c++ compiler. Will investigate.... Chris Am 09.07.19, 14:12 schrieb "Markus Sommer" <[email protected]>: Hi Chris, it looks like there's a problem with _declspec. It's probably nothing big. Unfortunately I won't be able to look it up until tonight. Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de<http://www.isb-fn.de> Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Dienstag, 9. Juli 2019 13:56 An: [email protected] Betreff: Re: [C++] Status Update? Hi Markus, Seems I have had progress. After noticing that I need a newer version of CMake in order to get real Boost 1.70 support, it took a while to get things running, but now it seems that it's mostly working. However almost at the end I'm getting this error, which I have absolutely no idea to what it means and if something's wrong with the code or my build: [ 94%] Building CXX object drivers/s7/CMakeFiles/plc4cpp-driver-s7.dir/src/main/cpp/org/apache/plc4x/cpp/s7/dllexports.cpp.obj C:\Temp\plc4x\plc4cpp\drivers\s7\src\main\cpp\org\apache\plc4x\cpp\s7\dllexports.cpp:34:10: error: expected constructor, destructor, or type conversion before '(' token _declspec(dllexport) PlcDriver* _CreatePlcDriverInstance() ^ drivers\s7\CMakeFiles\plc4cpp-driver-s7.dir\build.make:62: recipe for target 'drivers/s7/CMakeFiles/plc4cpp-driver-s7.dir/src/main/cpp/org/apache/plc4x/cpp/s7/dllexports.cpp.obj' failed mingw32-make.exe[2]: *** [drivers/s7/CMakeFiles/plc4cpp-driver-s7.dir/src/main/cpp/org/apache/plc4x/cpp/s7/dllexports.cpp.obj] Error 1 CMakeFiles\Makefile2:256: recipe for target 'drivers/s7/CMakeFiles/plc4cpp-driver-s7.dir/all' failed mingw32-make.exe[1]: *** [drivers/s7/CMakeFiles/plc4cpp-driver-s7.dir/all] Error 2 Makefile:82: recipe for target 'all' failed mingw32-make.exe: *** [all] Error 2 [INFO] ------------------------------------------------------------------------ [INFO] BUILD FAILURE [INFO] ------------------------------------------------------------------------ Chris Am 08.07.19, 20:52 schrieb "Christofer Dutz" <[email protected]>: Ok, Think im taking down the root of the problems. At first I found out, no matter how I configured things, the build always used the installed version of boost. Now that I forced it to locally use it, it started failing. Today I found out that the version of cmake the plugin is using inbox supports boost till 1.63.0 and simply doesn't find newer versions. So I'm currently trying to update the build to use newer cmake version. Hope I'll fix that tomorrow as it didn't feel right working around the tooling. The more I work in this c++ environment, the more I feel a deep love for the Java ecosystem 😘 So far the update... Chris Holen Sie sichOutlook für Android<https://aka.ms/ghei36> ________________________________ From: Markus Sommer <[email protected]> Sent: Friday, July 5, 2019 3:38:03 PM To: [email protected] Subject: AW: [C++] Status Update? Hi Chris, many thanks for your work :-). Best regards Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de<http://www.isb-fn.de> Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Freitag, 5. Juli 2019 15:04 An: [email protected] Betreff: Re: [C++] Status Update? Hi Markus and others, I invested about a day in cleaning up the C++ build: - I removed all the other pom files and moved the boost build down to the root of the plc4cpp. - I added some missing file headers in two files - I managed to somehow get the build to work with regards to the boost include directories (Seems the cmake find_packages function doesn't correctly set) - I managed to force CMake to use the boost version we provide (It seems to have always found the installed version on my system and never used the one I wanted it to) - I managed to spot the bug that was causing build errors in the log module (Turned out it was simple usage of backslashes instead of forward slashes) Now it think the whole "Boost_INCLUDE_DIRS" thing doesn't seem to work on my machine. I am providing Environment variables as hints and cmake claims to have found the directoeis but in contrast ot the documentation the variable "Boost_INCLUDE_DIRS" is never defined. I changed the CMakeLists.txt to reference another Environment variable "BOOST_INCLUDEDIR" which seems to work, but you might have to ensure to provide it too. Chris Am 04.07.19, 16:49 schrieb "Markus Sommer" <[email protected]>: Hello, Chris, I've already seen the MAVEN always fetches the current state of Boost, which is a problem when changing interfaces. Here we have to download a defined version of Boost when loading Boost from Maven and then CMAKE finds the Boost Library and can set the variables correctly. Currently, setting the variables for Boost is set in the parent CMAKE and all sub CMAKE then use the variables. Best Regards Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de<http://www.isb-fn.de> Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Donnerstag, 4. Juli 2019 16:38 An: [email protected] Betreff: Re: [C++] Status Update? Hi Markus, I just pulled your latest changes, uncommented the parts that will not compile on Mac / Linux and am currently working on the build / maven stuff. I did change that I pulled the boost stuff down to the root and updated the BOOST_ROOT to: set(BOOST_ROOT "./target/boost_1_70_0/") Most seems to be working with this ... but still the logger module is complaining to not be able to find the boost includes ... I'm currently trying to find out what's wrong there. Regarding the loading of modules ... doesn't boost have some abstraction for this? I thought Boost is there to abstract from exactly these types of differences between the OSes. Chris Am 04.07.19, 16:19 schrieb "Markus Sommer" <[email protected]>: Hey Chris, Meanwhile a S7.dll can already be created for Windows and with the example hello plc4x we can successfully link DLL into the exe. Auf Basis das Connectiontyps S7, ModBus, etc. wird die DLL dynamisch nachgeladen. The connection is implemented. The next steps are: Generation via Maven/CMAKE Reporting telegrams to the reassigned levels vie Signal2 Implementation of telegrams with new procedure of automatic generation Currently, however, the driver can only be compiled with CMAKE under Windows. Why? The interface for loading the DLL is different between Windows and Linux. As part of the Maven customizations, we would also make the customizations for Linux. So far we have created the CMAKE files for creating the PLC4CPP driver. We haven't deleted the Maven parts yet. This would be the procedure for generating the complete driver: Starting the BOOST Library generation process via Maven(as now) Start Plc4x CMAKE process via MAVEN We haven't done that much with Maven yet, we still have to get used to it. Chris might be able to do the integration in a Telko. Best regards Markus Freundliche Grüße Markus Sommer Geschäftsführer isb innovative software businesses GmbH Otto-Lilienthal-Strasse 2 D - 88046 Friedrichshafen Tel.: +49 (0) 7541 3834-14 Mob: +49 (0) 171 537 8437 Fax: +49 (0) 7541 3834-20 E-Mail: [email protected] Web: www.isb-fn.de<http://www.isb-fn.de> Geschäftsführer: Markus Sommer, Thomas Zeler Sitz: Friedrichshafen Registergericht: Amtsgericht Ulm HRB-Nr. 631624 Important Note: This e-mail and any attachments are confidential, may contain trade secrets and may well also be legally privileged or otherwise protected from disclosure. If you have received it in error, you are on notice of its status. Please notify us immediately by reply e-mail and then delete his e-mail and any attachment from your system. If you are not the intended recipient please understand that you must not copy this e-mail or any attachments or disclose the contents to any other person. Thank you. -----Ursprüngliche Nachricht----- Von: Christofer Dutz <[email protected]> Gesendet: Donnerstag, 4. Juli 2019 15:28 An: [email protected] Betreff: [C++] Status Update? Hi All and especially Markus, I just wanted to hear what the C++ part is up to. I can see a commit every now and then in the s7-cpp branch, but not much discussion (Actually none at all). Would be cool, if you could give us all an update to what you’re up to. Also would I like to tackle the problem of being able to fully build C++ from the maven build. We had discussed the options a while ago, but nothing has happened so far. I think the way I understood our last state we decided to: * Get rid of all the intermediate pom.xml files and all the Maven packing and unpackeing of stuff * Concentrate fully on starting one CMake build in the plc4cpp root pom and have that build all modules * We do the packaging using the maven assembly plugin which is also defined in the plc4cpp module So I have a few questions: * How do we integrate the downloading and building of boost in this szenario? * Do we simply copy the boos build stuff to execute before running the CMake build in the plc4cpp module? * Do we add something to the CMake build that takes care of this? * Currently can’t come up with the second question ;-) Chris
