compiling gettingstarted for wiringPi fails #669
Comments
2bndy5
changed the title
|
various answers (about I'm going to disable compiling examples for wiringPi in the workflow until this is solved or deprecated. Personally I'm erring towards deprecating it since its not a library provided by the nRF24 org, but I'm just not familiar/versed enough with wiringPi to address this on my lonesome. |
|
So it would seem "wiringPi.h" isn't providing the #ifdef RF24_WIRINGPI
#include "wiringPi.h"
#endifBut, it didn't help any. I still haven't looked into source code of the fork of an unofficial mirror of the deprecated wiringPi repo that the workflow is using... This seems like the next logical step since the |
* don't hide EN_ACK_PAY reqs; adjust examples & docs * removed duplicated calls in examples/tests * too many examples * update docs about reUseTx() * deprecate enableDynamicAck() * clarify reUseTx() * typo in func ref * append setAutoAck() about pipe 0 * consistent casing in "auto-ack" * fix typo * revert this "see also" changes * use ack_payloads_enabled to full potential * undeprecate flush_rx() * isPVariant return internal var; default dyn_pl on addresses the following issues: - #401 - #641 - #660 * oops, readdress #401 #641 #660 * address #640 * deprecate isAckPayloadAvailable() #664 * try to ignore return value; trigger forked actions attempting to address #636 * this is why I can't have nice things * retrying to ignore return value; fix init order * add CI for avr compiling examples * build doxygen with latest release ver number * revert doxyfile, fix doxygen.yml, make2 = install * build examples changes, doxygen.yml bad indent * remove unnecessary pipe chars in doxygen.yml * linux CI changed, pde->ino files * platform specific examples; fix WiringPi build * fix using void return value * remove gemma & lilypad; fix certain example vars * wiringPi may be broken #669 * 8 paragraphs that say keep auto-ack enabled #667 * reverted examples folders; exclude maple & 3pin * exclude didn't work * trigger arduino build on workflow changes * explicit examples' paths * no wildcards in explicit paths * no quotes in yaml sequence * try using a pipe char * arduino build "needs" A-style check * applied arduino IDE's auto formatting * remove format checker * fine let the example terminate * all official arduino cores to build_arduino.yml * backup examples; new GettingStarted * no explicit examples; compile all * new InterruptConfigure example * // about when IRQ flags get cleared; & a fix typo * new StreamingData & MulticeiverDemo examples * reduce streamData compilation size * tested GetStarted, AckPayloads, Streaming * IRQ & Multicceiver examples tested * undo #646; add broken FakeBLE example * fix arduino build CI * forgot to use pipe char * typo * 2nd job should run on ubuntu-latest * fqbn arch-type is wrong?! * uncomment #define IRQ_PIN from testing artifacts * use SpenceKonde core released zip file * use SpenceKonde board definitions * use correct vendor name?! * use lower case?! * platform name should match the fqbn * what am I doing wrong? use SpenceKonde attiny core * under "platforms" not "with" * yml formatted "platforms" options * "with: platforms:" * source-path uses relative path * use trailing / in source-path * clear advice from per1234 about yml syntax * fixing source-path and version number * use direct download link to SpenceKonde release * adjust source-path again * no source-path * redo-ing #646 * timingSearch3pin includes stdio * don't ignore rf24_ATTiny examples * separate job for separate ATTiny examples * timingSearch3pin uses redefined spi funcs * drop testing on attiny43, install core from json * keywords updated removed duplicate examples * CallResponse was duplicate; (+) ManualAck example * doc updates #671 tested new manAck example * FakeBLE crc calc not checking out * add per1234's action to check formatting * did not find expected alphanumeric char?! * add manualAcknowledgements example to docs * don't remove comment block prefix * restored comment prefix in old examples also * less confusing comments about addresses * clarify docs on return value for all write-related * remove broken fakeBLE example * forgot to remove fakeBLE from arduino workflow * test deploy to gh-pages * bad yml format * something's wrong with the doxygen build CI * lack of new line in end of doxyfile * remove debug prompt; deploy on publish released * deploy on any release event * fix main page useful links * only clear RX_DR flag in read() * typo * doc updates; new python examples * new gettingstarted example for linux * add micros() to applicable utilities compatiblity * define micros() in arch_config.h * printDetails() in linux * pipe number not printing correctly * oops it was a non-printable char * avoid keyboard interrupt * build wiringPi examples also * linux build CI for SPIDEV * ignore return value in utilty/SPIDEV/interrupt.c * trying rf24_micros() * ammend example to use rf24_micros() * let micros() be defined in examples for linux * need cstdio * ported arduino examples to linux (& python) * oops * should probably declare i huh * copy n paste much? * don't use stoi for a char * typo, expose digitalRead(), ifdef RF24_WIRINGPI * ifdef WIRINGPI didn't help, fix inline condition * testing round 1; fix pyRf24/setup.py * bug fixes in py examples prompt fixes in linux * fix irqConf for arduino, rename stream_data.py * copy n paste artifact * debugging streamingData * testing round 2 * oops * timing output changes; printDetails() also * debugging IRQ configure example * debugging IRQ_config.py also * declare timer stuff for micros in manACK example * typo * doc read() for python; add flush_rx() to py wrap * remove cap on how many bytes to read in python!!! * don't cap bytes read() from RX FIFO in C++ * supposed to be decode() not encode() * fix typos; better python doc about read() * py wrapper omits len args for buf-related funcs * fix streaming_data.py master() * duplicate prompt in ackPay.py * prompt format changes * big commit - reverted defaults about Dynamic Payloads and Dynamic ACKs per @TMRh20 suggestions about Si24R1 in #658. This reverts #660 & #661 - ammended docs & all examples about above suggestions - re-wrote InterruptConfigure examples to use attachInterrupt(); py example uses RPi.GPIO equivalant. - removed MKR-based boards from Arduino build CI as pin 2 on those boards don't support interrupt requests, and InterruptConfigure.ino uses pin 2 for exactly that. - added building of python wrapper to Linux workflow; it now also runs pylint on python examples - removed my attempt to port digitalRead() to various Linux drivers as it is no longer needed in the interruptConfigure.cpp example. - addressed #674 about RX_PW_Px registers manipulation. - remove and ignore my VSCode folder as it is machine-specific - removed contradicting "examples_linux" ignores and added "examples_linux/__pycache__" folder - RF24::begin() now brute forces a soft reset. It does not reset addresses set to the pipes, but it does close all pipes. This helps address #401 - fixed some typos/copy-n-paste issues in the docs. - added instructions on how to run the python example scripts * Linux build CI use pip3 to install RPi.GPIO * workflow uses python 3.7 x86 * workflow uses python 3.x x86 * python needs x32 called in workflow * fix address assignment in linux examples * don't check python examples in workflow * define INPUT in RPi-BCM driver * remove GPIO.cleanup() from py examples * payloadSize is an attribute in python * fix python examples about getPLsize() * use RuntimeError in python examples when begin() is false * testing round 3 * streamData uses write(); python examples need 3.7+ * fix ellapsedTime in streamingData * status byte saved as private member addresses #678 #679 #677 removed comments in build_linux.yml about building/testing the python wrapper added a fail fast prompt to the streaming examples when failures >= 100 * simplify logic about _is_p_variant * testing round 4 & address #680 * fix irq examples * fix compilation errors for IRQ examples * fix irq example for linux * tweak IRQ py example; writeAckPayload returns bool * final tweaks to IRQ examples; some doc updates * leave closeReadingPipe(0) in startListening() * arduino examples work address 2 new issues 1. #687 added printPrettyDetails(); updated python wrapper and keywords.txt 2. #686 centralizing all SPI access to write_*() and read_*() functions Also reverted changes to gitignore about examples_linux folder while adding/ignoring some things about building the python wrapper (venv, *.pyc, __pycache__) * remove volatile from radio declaration * datasheets; fix man_ack.py example updated comment proceeding radio object in python examples documented enums and linked references to them. Also linked references in COMMON_ISSUES.md & main page's changelog. * ATTiny examples should work on SpenceKonde core * exclude ATTinyx313 from timingSearch3pin example * oops, bad indent * Astyle indent for #ifdef is 0 * (+) CLI args to examples. pwr down radio on ctrl+c * fail if no valid arg is specified * fix parsing args in python examples * make py examples behave like c++ examples * found a bug in py examples' -r arg * testing round 5 * adjust manpage descriptions * don't use signal.h in cpp examples * del cli args from all linux examples except 2 * edit doxyfile OUTPUT_DIR * only run doxygen on master branch * redirect doc links to nRF24 org * missed a link * remove use of timeout bool in manAck.ino * self-reviewed PR changes * ATTinyCore adjustments - use macros from SpenceKonde ATTinyCore - add macros for CSN settling times for when using the ATTinyCore * some more self-reviewed PR changes * exclude root README from doxygen output * exclude GPIO & SPI classes from doc'd classes * add Arduino & Linux build badges to readme * Change link to docs in readme
2bndy5
changed the title
|
I think the linker when cross-compiling with arm-linux-gnueabihf compiler (like we do in the GitHub action workflow) doesn't automatically search /usr/local/include folder for installed lib headers. However, it seems the linker might find the correct symlink installed to /usr/local/lib... I'm still learning as I go here. |
|
OK! I just got the CMake CI job to successfully link to WiringPi & MRAA libs. Examples are also compiling successfully too 🥇 . I'm leaving this issue open until we officially deprecate the traditional build system (using ps - I did notice that MRAA and WiringPi drivers are not compatible with the interruptConfigure.cpp example (as it is currently written). So, the CMake for Linux implementation skips that example accordingly. Also, just noting that wiringPi requires executables to be additionally linked to libcrypt and librt ( |
|
Setting aside what I said about executables required linked libs for wiringPi (which may be only necessary when cross-compiling on non-arm host machines), I tried patching the examples_linux/Makefile to conditionally include interruptConfigure.cpp if wiringPi or MRAA drivers are used. I humbly admit I got stopped by seemingly common (& extremely vague) makefile syntax errors. This may be the best way to fix this issue once I can get wiringPi & MRAA building, installing, and linking to the RF24 library successfully in the workflow (they are currently disabled as of this writing). When compiling locally on a RPi, I expect that the current builds with wiringPi output an error about the interruptConfigure.cpp example. Because it's the last example that the Makefile compiles, all other examples should still compile fine. |
|
Closing this as wiringPi support is now frozen (as in no further maintenance will be provided). Please use pigpio for IRQ support (more preferably the SPIDEV or RPi drivers). |
|
I finally solved this one. The fix for this is in char-dev-irq branch RF24/examples_linux/Makefile.examples Lines 20 to 22 in bd48742 |
2bndy5
changed the title
In an effort to extend the same CI strategy from using Arduino CLI to the Linux platform, I've added an extra step to the build_linux.yml workflow to go further and compile the examples in the "examples_linux" folder. This idea first came from #642.
compiling the examples with WiringPi fails with the following error message (copied from my branch's workflow log):
I have a feeling this is actually a simple fix, and it may even be specific to the the workflow (RF24 library builds fine). Nonetheless, I fear I'm not well enough equipped to address this.
The text was updated successfully, but these errors were encountered: