Відео

Iyi calismalar, direk beaglebone uzerinde de derleyebilirsin. Fakat cross development yapmak istiyorsan, benim onerim sanal makina yada docker gibi katmanlari kullanmadir. Bu sekilde isletim sisteminden bagimsiz bir sekilde derlersin. Windowsda direk cross derleyip linux uzerinde calistirmak gereksiz olabilir. Senin yerinde olsam en son paylastigim docker ile cross derleme videosuna bakardim. Ordaki docker dosylarini modifiye eder gerekli sonuca ulasirdim. Qt icin cozumu benim yaptigim icerikten ulasirsin gibi duruyor bir kac deneme ile. beaglebone zaten debian based olmasi lazim cok bir fark yok. Opencv icin ise, ilk basit bir kac opencv ornegini cross olarak derle cmake ile sonra beagleboneda calistigini gor. Ardindan zaten docker yada sanal makina kullanarak qt ile olan adimlari birlestirirsin. CMAKE_TOOLCHAIN_FILE dosyasi toolchain ile alakali katmanlari icermeli sanki, opencv cmakelists.txt de cagirilabilir. Opencv yi qt gibi dusun, sadece so(shared object) dosyalari var ve onlari cmake ile bind etmen lazim. (bide header lar tabi) :)

@@ulasdikme7307 Buna bakacağım, C++ tarafı ikinci uygulamam olacakta , şu anda spyder idesinde pythonda yazdığım logitech kamerasına ulaşan opencv kodu, beaglebone'da çalıştıramıyorum :). Gstreamer hatası alıyorum ayrıca kamera verisi alamdığından timeout oluşuyor ama kameraya ulaşıyor ledin yanmasından anlıyorum. Önce bu tarafı halletmem önemli.

If image is same then yes☺️

Hi, you need to update dockerfile to cross compile mqtt. Then it should work.

Hi, Linux to windows? I recommend you to use virtual machine or docker instead of windows as a os. so you can compile between different hardwares with linux.

Qtcreator ile yapmaya calisiyorsan olmaz video da vscode kullaniliyor, qmake deil cmake kullanman lazim. qmake icinde cmake komutu kosamazsin.

@@ulasdikme7307 eyvallah baskan bende creator la kurarken cmake ile kurdum cmake file var yani. En iyi crossfile hangisi misal creator ile kursam yapsam komple projec ti sonra bitirmek

@@R1FighterR1 ben cmake ve vscode u oneririm, ama hizli bitirryin diyorsan debug yapmayi bosver cmake ile komut satirindan derle istedigin ide yide development olarak kullan. Videoda anlattigim gibide yapabilirsin qtcreator nie kullanmak istiyorsun?

Thank you, It is kind of a magic for me too :) If you want to share the script to build whole qt do not hesitate it. I never tried whole build and I do not think that I will have enough time to try. If you share it other people can take as a reference.

@@ulasdikme7307Thank you for the encouragement! I've made changes to the workspace creating different folders in order to test compatibility with RPI Zero 2 and different versions of Qt. Initially, I tried RPI4 running bookworm 64 lite, using with eglfs instead of xcb. So, once the container generates the tar file, I need to manually install some additional libraries. Let me conduct a few more tests to clarify. Give me a couple of days and II'll get back to you!

@@edulum1318 Sure take your time 👍

orinCroosTool.cmake cmake_minimum_required(VERSION 3.10) include_guard(GLOBAL) # Set the system name and processor for cross-compilation set(CMAKE_SYSTEM_NAME Linux) set(CMAKE_SYSTEM_PROCESSOR arm) # Set the target sysroot and architecture set(TARGET_SYSROOT /home/yao/orin/sysroot) set(TARGET_ARCHITECTURE aarch64-linux-gnu) set(CMAKE_SYSROOT ${TARGET_SYSROOT}) set(CMAKE_FIND_ROOT_PATH ${TARGET_SYSROOT}) set(CMAKE_INCLUDE_PATH ${TARGET_SYSROOT}/usr/include) set(CMAKE_INCLUDE_PATH ${TARGET_SYSROOT}/usr/include/aarch64-linux-gnu) set(CMAKE_LIBRARY_PATH ${TARGET_SYSROOT}/usr/lib) # Add library directories list(APPEND CMAKE_LIBRARY_PATH "${TARGET_SYSROOT}/usr/lib") list(APPEND CMAKE_LIBRARY_PATH "${TARGET_SYSROOT}/lib/aarch64-linux-gnu") list(APPEND CMAKE_LIBRARY_PATH "${TARGET_SYSROOT}/usr/lib/aarch64-linux-gnu") set(CMAKE_INSTALL_RPATH "${TARGET_SYSROOT}/lib/aarch64-linux-gnu") list(APPEND CMAKE_RPATH_LINK_PATH "${TARGET_SYSROOT}/lib/aarch64-linux-gnu") # Add RPATH link directories list(APPEND CMAKE_RPATH_LINK_PATH "${TARGET_SYSROOT}/usr/lib") list(APPEND CMAKE_RPATH_LINK_PATH "${TARGET_SYSROOT}/usr/lib/aarch64-linux-gnu") list(APPEND CMAKE_RPATH_LINK_PATH "${TARGET_SYSROOT}/lib/aarch64-linux-gnu") # Set paths for CUDA libraries set(CUDA_INCLUDE_DIR ${TARGET_SYSROOT}/usr/local/cuda/include) set(CUDA_LIB_DIR ${TARGET_SYSROOT}/usr/local/cuda/lib64) # Configure the pkg-config environment variables set(ENV{PKG_CONFIG_PATH} "$ENV{PKG_CONFIG_PATH}:${CMAKE_SYSROOT}/usr/lib/${TARGET_ARCHITECTURE}/pkgconfig") set(ENV{PKG_CONFIG_LIBDIR} "${TARGET_SYSROOT}/usr/lib/pkgconfig:${TARGET_SYSROOT}/usr/share/pkgconfig:${CMAKE_SYSROOT}/usr/lib/${TARGET_ARCHITECTURE}/pkgconfig:${CMAKE_SYSROOT}/usr/lib/pkgconfig") set(ENV{PKG_CONFIG_SYSROOT_DIR} "${CMAKE_SYSROOT}") # Set the C and C++ compilers set(tools /home/yao/orin/gcc-linaro-7.5.0-2019.12-x86_64_aarch64-linux-gnu) set(CMAKE_C_COMPILER ${tools}/bin/${TARGET_ARCHITECTURE}-gcc) set(CMAKE_CXX_COMPILER ${tools}/bin/${TARGET_ARCHITECTURE}-g++) # Define additional compiler flags set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -isystem=${TARGET_SYSROOT}/usr/include -isystem=${TARGET_SYSROOT}/usr/local/include -isystem=${TARGET_SYSROOT}/usr/include/${TARGET_ARCHITECTURE}") set(CMAKE_CXX_FLAGS "${CMAKE_C_FLAGS}") # Configure CMake find root path modes set(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER) set(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY) set(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY) set(CMAKE_FIND_ROOT_PATH_MODE_PACKAGE ONLY) # Set the install RPATH use and build RPATH set(CMAKE_INSTALL_RPATH_USE_LINK_PATH TRUE) set(CMAKE_BUILD_RPATH ${TARGET_SYSROOT}) # Append to CMake library and prefix paths list(APPEND CMAKE_LIBRARY_PATH ${CMAKE_SYSROOT}/usr/lib/${TARGET_ARCHITECTURE}) list(APPEND CMAKE_PREFIX_PATH "${CMAKE_SYSROOT}/usr/lib/${TARGET_ARCHITECTURE}/cmake")

CMakeLists.txt cmake_minimum_required(VERSION 3.10) project(yolov5-tensorrt VERSION 0.1 DESCRIPTION "Real-time object detection with YOLOv5 and TensorRT") set(CMAKE_CXX_STANDARD 17) set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O2 -g -Wall -Wextra -Wno-deprecated -fPIC") set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -g -pthread") set(CMAKE_INCLUDE_CURRENT_DIR ON) set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}) if(NOT DEFINED CMAKE_LIBRARY_OUTPUT_DIRECTORY) set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}) endif() find_package(OpenCV REQUIRED) # find_package(CUDA REQUIRED) include_directories(include) file(GLOB SOURCES "src/*.cc") add_executable(camera_detect main_camera.cc ${SOURCES} ) target_include_directories(camera_detect PUBLIC ${OpenCV_INCLUDE_DIRS} ${CUDA_INCLUDE_DIR} ) target_link_libraries(camera_detect nvinfer nvonnxparser ${CUDA_LIB_DIR} ${OpenCV_LIBRARIES} )

What is the error?

there is a build.log file in the container, you can find more info.

@@ulasdikme7307 how we can open that build.log for the container where we cant build the container

@@jordantrajkov9549 docker cp tmpbuild:/build.log ./build.log it is explained in the github github.com/PhysicsX/QTonRaspberryPi?tab=readme-ov-file#debugging-of-compilation

@@ulasdikme7307 okay thanks mate

hahahah 😁 Turkce de yaparim siz abone olun. Sorun yok. Bu arada yeni docker ile bir tutorial yaptim onu ornek alirsan daha iyi olur bu eski.

@@ulasdikme7307 abone oldum. cevap vereceğini hiç düşünmemiştim :) docker de bilmiyorum ki hiç kullanmadım. şimdi ona da bulaşmayayım bu ara çok şey öğrendim-öğreniyorum kafam çorba oldu. Yani benim gibi ingilizce özürlüleri için lütfen Türkçe de yapın.

Docker lisi daha kolay. Videoyu izle sadece komutlari kos. Tum bagimliliklari ben hallettim sadece komutlari kosarsan bir kac saat sonra qt compile olmus bir sekilde kullanima hazir olarak bulumacak. Github da komutlar var. Docker dedigim sey sanal makina yerine kucuk izolasyon yaratiyor, rasp a falan ihtiyacin yok tum derlemeyi host da yapabiliosun acayip kolaylik. Takildigin yer olursa sor cevaplarim. Github linki video nun altinda. Turkce cok izlenmiyor dedim bari ingilizce yapayim oda cok izlenmedi hos 😁 Ama dusunuyorum zamanim olursa genel bir video yapacagim turkce

Hey thanks, follow the latest one, I have been keeping this up to date.

@@ulasdikme7307 which one of them please ? im facing the same problem of "no such file or directory" (cant find the /opt/vc)

@@mayssaknani You can try this video. There are steps in the link below the video.(github link) With docker it is easy, but you need to update docker files to compile opecnv if you need it.

I may have been a bit silly and added these to the dockerfile instead of the DockerFileRasp. Will update if that fixes things.

EDIT: Issue solved: Putting libmodbus-dev \ and libmodbus5 \ into the run sections of the DockerFile and DockerFileRasp seems to fix the errors in the earlier steps. I had a second error related to linking the library during the final step of making the executable the solution was to update where my custom modbus module was looking for it. I put this in the CMAKELISTS files referencing the modbus libraries, it points to where the actual library files ended up in the build image: target_link_libraries(ModbusModule PRIVATE Qt6::Qml /build/sysroot/usr/lib/aarch64-linux-gnu/libmodbus.so)

If you add these libraries in dockerfile, then they are installed for ubuntu which is x86 probably, but you need them for rasp. I think binding will be a problem in this case.

Hi, I just tested link is working, probably you may have vpn or network configuration which does not allow you to access these links?

i have tried with different internet connections but it doesn’t change,i think the error is about missing command of “wget” when the RUN line of tar files in DockerFile.Is there a way to fix it anyway ? Can I download manually these .tar files and implement the docker build ?

@user-gw3wu9vq4x I just tested and it is working. You can download them in the same directory where dockerfile exists. Then using COPY command it is possible to copy them to image. But as I told you, if you have problem with internet connection, you can see another issues.

Hello Eriton, I hope you are fine, If you have the HelloQt6 then you compiled the qt successfully, good! now you need to send qt-pi-binaries.tar.gz to target and extract it and you export LD_LIBRARY_PATH on the target like in the video or in the link below the video. Have you did that?

@@ulasdikme7307 Thanks for helping! But I sent qt-pi-binaries.tar.gz to target and extract it and export LD_LIBRARY_PATH. But I show this error.

Could you please share with me output of echo $LD_LIBRARY_PATH and sudo find /usr/local -name libQt6Core.so.6 and ldd HelloQt6, lets check together

@@ulasdikme7307 Sorry, this tuturial is tar -xvf qt-pi-binaries.tar.gz -C /usr/local, I change to tar -xvf qt-pi-binaries.tar.gz -C /usr/local/qt6 Works fine now! Thanks for helping!!!

Enjoy.. Do not forget to subscribe, share and like 😇

Thanks, I have been working on vscode for debugging but It is not working as I expected. I would like to create another video for it if I address the issues. For qtcreator you need to extract syroot and cross toolchain and qt libraries(these are available in the container already as tar file), then you should give the path in qtcreator when you add new device and compiler. You need to run qtcreator on host for remote debugging. Do not forget to subscribe and share ! :)

@@ulasdikme7307 Sure. Thanks for the promt reply. Eagerly wait for your vs code and qt creator video. Keep up your good work , very helpful. I will try this approach this weekend. I have few follow up questions. If could clarify that would be great. Thanks kindly. 1. For raspberry pi hardware how so we match QEMU image, do you flash the same image as QEMU if yes can you please explain how ?. 2. Can i install QT creator on pi desktop directly and configure the ql libraries after they the compiled libs are sent to the raspberry pi hardware ? basically after cross compilation of qt lib for raspberry pi , can we just use only raspberry pi desktop to write applications?

@@audiojog 1. This can be selected via dockerfile FROM --platform=linux/arm64 arm64v8/debian:bookworm as rasp. If you check the docker file for raspberry pi, then you can see bookworm type. But becareful, as I explained in the video, because of different hardware it is not possible to match exact os, there will be difference everytime. Related hardware differences(precompiled binaries) can be found directly from rasp repository. It is explained in the video. When I compiled the Qt without docker, I see some differences in the configuration from eglfs perspective with compilation using docker. it is becuase I missed some precompiled libraries probably. But it is fine for now. I will check it later. 2. I do not recommend using any ide or compilation on target device. But yes if you want, you can run it on raspberry pi. It will be slow I think. You have all libraries and related dependencies already to compile and start. Therefore it is against the nature of cross compilation. :)

Hi, you can still compile it with vm, but yes docker takes muc longer time you need more ram. But it is easier than solution with vm as long as you have faster computer. Btw I did not show how to debug application with docker? I just showed how to compile? You can still build your development environment with vm using tar files which are produced by docker. If you see last lines in the dockerfile you will see 3 tar files, if you extract them in vm you will have old vm based development environment.

I published new tutorial about how to cross compile with docker. Isolation is good. Maybe it can help. Check my last video.!

Hi, I made another tutorial abouit it with docker ua-cam.com/video/5XvQ_fLuBX0/v-deo.htmlsi=iiR92LvUI-0D_cKP It is quite new and easy! I tested with rasp 4 but I am sure it will work for rasp 5. I think they use same image. Give it a try!

@@ulasdikme7307 very cool ! will test it on my RPI 4, waiting for the RPI 5 adn will test on it too! thx and great work!

This tutorial is quite old, because of toolchain you can see some issues, here it is updated version github.com/PhysicsX/QTonRaspberryPi/tree/main/QtRaspberryPi6.6.1 and also I created a very easy way with docker, I recommend you to check this page also github.com/PhysicsX/QTonRaspberryPi/tree/main/QtRaspberryPiWithDocker6.6.1. I will make a video about it soon.

@@ulasdikme7307 also makes sense to bring over etc/alternatives over with rsync.. I have been weeding though most of the pkg-config pc today.. up to the glib needing pkg and pkg neeeding glib.. anyway I found something strange on bookworm, Pkg-config exists glib returns a 0 on the pi itself .. and that is where it stops on cross compile build.. in the morning I will get out my snow shoes and start walking up that mountain, thanks for you input..

Hi, yes it is quite old. But the way is correct. You need to change toolchain. github.com/PhysicsX/QTonRaspberryPi/tree/main/QtRaspberryPi6.6.1 in this link there is new version, ( while copying the commands, i could made a mistake, but it is working) so if you check carefully, I compiled the toolchain. If you follow the steps, it will work. Currently I am thinking to create a new content about it but lets see.

Thanks, I will try your suggestion, I've already spent 3 days of working for the moment...@@ulasdikme7307

Hello, I've tried your new tutorial, but I can't compile toolchain - I'm using a raspi 3 - I've mistake with headers_install (not found command) - Can I still use Linearo Tool chain instead ?@@ulasdikme7307

I shared a new video, with docker you should be able to compile it if you follow the steps, please watch the new one also. Which version of toolchain you use? I tested with rasp 4 there can be differences with rasp 3.

Thanks, I'm going to try your new tutorial .I used rpi-gcc-8.3.0 @@ulasdikme7307

Tutorial is quite old, so it needs update. There is an official tutorial on qt website. I recommend you to check it. If I find time, I will update it. btw I like your playlist.

@ulasdikme7307 I was able to download latest qtbase on windows and transfer it to VM, otherwise links didn't respond on VM. Hope I'll get eventually to cross compile "hello world" 🤞. Anyway thanks for explaining the steps in the tutorial, it helps!