How did adding /usr/lib/ to my DYLD_LIBRARY_PATH cause this dynamic linking problem on my MacOS Ventura M1 MBP? I had set export DYLD_LIBRARY_PATH="...:/usr/lib:..." in my .bash_profile.

$ xz --version
dyld: Symbol not found: _lzma_cputhreads
  Referenced from:  /opt/homebrew/Cellar/xz/5.4.3/bin/xz
  Expected in:      /usr/lib/liblzma.5.dylib
Abort trap: 6
$ otool -L /opt/homebrew/Cellar/xz/5.4.3/bin/xz
/opt/homebrew/Cellar/xz/5.4.3/bin/xz:
	/opt/homebrew/Cellar/xz/5.4.3/lib/liblzma.5.dylib (compatibility version 10.0.0, current version 10.3.0)
	/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1319.100.3)

Unsetting this variable fixed my immediate problem.

$ DYLD_LIBRARY_PATH="" xz --version
xz (XZ Utils) 5.4.3
liblzma 5.4.3

However, looking in /usr/lib, I cannot find why my system thinks there was a version of liblzma in /usr/lib at all.

$ ls /usr/lib | grep 'liblzma'
$ find /usr/lib -name '*liblzma*'
$

I could understand seeing these symptoms if there was a version of liblzma in that directory, but there does not appear to be. /usr/lib was neither the first nor the last path in that list, so I don't understand why the failure would manifest that way in particular. So why did having that directory on my DYLD_LIBRARY_PATH cause my system trying to link to a dylib in that directory, even though said dylib does not exist at that location?

I am writing c++ code, and using cmake to compile the application. I want to add the dylib (libraries) to the Frameworks folder. Macdeployqt does not find them all. I have written my own python script to try and find most, but a few still elude me. Here is a short list of what I still can't find: Dependency Not Found! /usr/lib/libz.1.dylib Dependency Not Found! /usr/lib/libiconv.2.dylib Dependency Not Found! /usr/lib/libbz2.1.0.dylib Dependency Not Found! /usr/lib/libexpat.1.dylib Dependency Not Found! /usr/lib/libresolv.9.dylib Dependency Not Found! /usr/lib/libxml2.2.dylib Dependency Not Found! /usr/lib/libsasl2.2.dylib I usually use brew to install (which I assume is in /usr/local/Cellar by brew --prefix and Cellar), I have so far recursively searched: /usr/local/opt/**/{lib_name} /usr/local/Cellar/**/{lib_name} Where the "**" can be any set of folders and names. Where else can I look, it is not in /usr/local/lib where it states it should be. Even a symlink would be useful. I need to find these files, how can I do that, even manually would be okay, but i'd prefer some script? Somehow the application can find these on my own computer when run, but otool lists me places (/usr/local/lib) where it cannot be found!

I'm trying to build an SFML sample without xcode (for reasons :-)). The prebuilt samples comes with:

pong> otool -L pong
pong:
	@rpath/sfml-audio.framework/Versions/2.5.1/sfml-audio (compatibility version 2.5.0, current version 2.5.1)
	@rpath/sfml-graphics.framework/Versions/2.5.1/sfml-graphics (compatibility version 2.5.0, current version 2.5.1)
	@rpath/sfml-window.framework/Versions/2.5.1/sfml-window (compatibility version 2.5.0, current version 2.5.1)
	@rpath/sfml-system.framework/Versions/2.5.1/sfml-system (compatibility version 2.5.0, current version 2.5.1)
	/usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 19.0.0)
	/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 159.1.0)

When I build it myself (build command follows), I rather obtain:

pong> otool -L a.out
a.out:
	@rpath/libsfml-system.2.5.dylib (compatibility version 2.5.0, current version 2.5.1)
	@rpath/libsfml-window.2.5.dylib (compatibility version 2.5.0, current version 2.5.1)
	@rpath/libsfml-graphics.2.5.dylib (compatibility version 2.5.0, current version 2.5.1)
	@rpath/libsfml-audio.2.5.dylib (compatibility version 2.5.0, current version 2.5.1)
	/usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 1300.36.0)
	/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1319.0.0)

This prevents my built version from running, as it fails to load the library. How would I tell g++ to generate the same @rpaths as those prebuilt? The build command I used is:

g++ Pong.cpp -I ../../include -L ../../lib/ -lsfml-system -lsfml-window -lsfml-graphics -lsfml-audio

In case this makes any difference, this is on Ventura.

I'm trying to figure out why a binary isn't linking a shared library. On Linux the standard way to do this is LD_DEBUG=libs that shows the path of all linked libraries, but this isn't producing any output for me. I've read several articles recommending this on macOS, so was it removed or is there some trick to using it? I'm running on Catalina and I've tried both bash and zsh shells. Example from macOS:

$ LD_DEBUG=libs uptime
19:18  up 10:06, 4 users, load averages: 2.00 2.21 2.04

Example from Linux:

$ LD_DEBUG=libs uptime
     25453:	find library=libprocps.so.7 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/libprocps.so.7
     25453:	
     25453:	find library=libdl.so.2 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/libdl.so.2
     25453:	
     25453:	find library=libc.so.6 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/libc.so.6
     25453:	
     25453:	find library=libsystemd.so.0 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/libsystemd.so.0
     25453:	
     25453:	find library=librt.so.1 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/librt.so.1
     25453:	
     25453:	find library=liblzma.so.5 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/liblzma.so.5
     25453:	
     25453:	find library=liblz4.so.1 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/liblz4.so.1
     25453:	
     25453:	find library=libgcrypt.so.20 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/libgcrypt.so.20
     25453:	
     25453:	find library=libpthread.so.0 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/libpthread.so.0
     25453:	
     25453:	find library=libgpg-error.so.0 ; searching
     25453:	 search cache=/etc/ld.so.cache
     25453:	  trying file=/lib/arm-linux-gnueabihf/libgpg-error.so.0
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/libpthread.so.0
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/libc.so.6
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/libgpg-error.so.0
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/libgcrypt.so.20
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/liblz4.so.1
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/liblzma.so.5
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/librt.so.1
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/libsystemd.so.0
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/libdl.so.2
     25453:	
     25453:	
     25453:	calling init: /lib/arm-linux-gnueabihf/libprocps.so.7
     25453:	
     25453:	
     25453:	calling init: /usr/lib/arm-linux-gnueabihf/libarmmem-v7l.so
     25453:	
     25453:	
     25453:	initialize program: uptime
     25453:	
     25453:	
     25453:	transferring control: uptime
     25453:	
 19:17:22 up 22:36,  3 users,  load average: 0.00, 0.00, 0.00

I'm thinking of using libmpv to provide video playback in my MPL 2 Flutter app. I know that Apple is weird about GPL stuff (and using GPL stuff in a greater work is awkward), so I was just wondering if I was actually allowed to. Here's the source code if it matters: https://github.com/UnicornsOnLSD/finamp

I recently read this post on netkas.org about running 32 bit command line apps on Catalina using Mojave 32+64bit libraries and frameworks, adding to the nvram boot-args the option no32exec=0. I extracted with Pacifist the /usr/lib, /System/Library/Frameworks and /System/Library/PrivateFrameworks folders from the Mojave installation app to ~/.mojave and then tried to run this 32 bit macOS app with the following terminal command: $ DYLD_ROOT_PATH=~/.mojave /Volumes/Lazesoft\ Mac\ Data\ Recovery/Lazesoft\ Mac\ Data\ Recovery.app/Contents/MacOS/Lazesoft\ Mac\ Data\ Recovery Here is what I get running it: 2020-06-08 17:26:18.630 Lazesoft Mac Data Recovery[1881:56237] /Volumes/Lazesoft Mac Data Recovery/Lazesoft Mac Data Recovery.app 2020-06-08 17:26:18.630 Lazesoft Mac Data Recovery[1881:56237] /Volumes/Lazesoft Mac Data Recovery/Lazesoft Mac Data Recovery.app/Contents/MacOS/ifsui 2020-06-08 17:26:18.632 Lazesoft Mac Data Recovery[1881:56237] Error Creating Initial Authorization: -60008 2020-06-08 17:26:18.632 Lazesoft Mac Data Recovery[1881:56237] AuthorizationExecuteWithPrivileges failed! Trying to run ifsui from the same app bundle I get the app to open in the dock but nothing appears on screen and this is what I get on the terminal: _RegisterApplication(), FAILED TO REGISTER PROCESS WITH CPS/CoreGraphics in WindowServer, err=-304 2020-06-08 17:21:38.555 ifsui[1836:52747] *** WARNING: Method userSpaceScaleFactor in class NSView is deprecated on 10.7 and later. It should not be used in new applications. Use convertRectToBacking: instead. I know I'm probably asking something that cannot really be done, but is there any way to make this workaround work with more than command-line programs?

The question is in the title, but anyway, let me explain it a bit more: The most accepted way for correctly defining the ***install name*** for a dylib in MacOS is by making it relative to the rpath. For example: otool -L ./LLVM/7.0.0/lib/libomp.dylib ./LLVM/7.0.0/lib/libomp.dylib: @rpath/libomp.dylib (compatibility version 5.0.0, current version 5.0.0) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1238.50.2) Here, libomp.dylib has @rpath/libomp.dylib as its ***install name***. So far so good. The problem is when you create an executable linked to such a dylib. Even if you pass the correct -L/path/to/libomp.dylib flag at link time, so that ld can **successfully** link the executable, then you try to execute it, and obviously you get the following: dyld: Library not loaded: @rpath/libomp.dylib Referenced from: mydumbexecutable Reason: image not found Abort trap: 6 This of course can be fixed by using install_name_tool on either the dylib (changing its ***install name*** so that it doesn't depend on the rpath, and linking the executable again, but this is not considered good practice), or, the recommended way, to use install_name_tool on the executable, adding to it the proper rpath so that the dylib can be found. But... just wondering... isn't there a flag in ld that automatically adds the rpath for you? I mean, if ld is able to link the executable because it **did** find the dylibs, why cannot automatically store the proper rpath in the executable? I understand this should be optional behaviour, as sometimes you prefer to define the rpaths yourself, but... a flag for doing it automatically would make my life a lot easier.

When I use ldd on Linux it tells me right away if something is missing, i.e.

ldd -d some_lib.so
    libexample.so => not found
    libex.2.so => not found
    libm.so.6 => /lib/libm.so.6 (0xf5860000)
    libdl.so.2 => /lib/libdl.so.2 (0xf785b000)
    libc.so.6 => /lib/libc.so.6 (0xf78a9000)
    /lib/ld-linux.so.2 (0x46b4c000)

On Mac OS there is no ldd and people recommend using otool -L, but that shows which shared libraries are expected, not weather or not they are there. Another approach would be to use dtruss, but it requires sudo, and it least in my experience the application behaves very differently when it is being dtrussed, and it fails for an entirely different reason (and before it gets to trying to load the troublesome .dylib). So if you suspect that a shared library is missing, what is the standard approach to finding which one it is on the Mac?

There are two Macs and one of them finds a certain dylib when I dlopen it and another one doesn't. Seems like the library search paths are different, but DYLD_LIBRARY_PATH and DYLD_FALLBACK_LIBRARY_PATH are empty on both machines. How can I see what paths dlopen searches?

I am trying to force the update of my dyld shared cache using the command sudo update_dyld_shared_cache -root / -force But after a while I see a bunch of errors like this, related to spindump: >update_dyld_shared_cache : Unable to unlink //var/db/spindump/OLDDSC-E2D77243-3FD9-32E5-AE64-3506904D5579: 1 Operation not permitted How do I fix this? I am on El Capitan.

I try to do brew install tmux it installs, but, probably due to previous failed and not fully removed non-brew attempts to install both it and libevent, when try to use tmux I get $ tmux dyld: Library not loaded: /usr/local/lib/libevent-2.0.5.dylib Referenced from: /usr/local/bin/tmux Reason: image not found Trace/BPT trap: 5 I tried to install libevent but I got: $ brew install libevent Warning: libevent-2.0.21 already installed, it's just not linked so I tried $ brew link libevent but I got Linking /usr/local/Cellar/libevent/2.0.21... Warning: Could not link libevent. Unlinking... Error: Could not symlink file: /usr/local/Cellar/libevent/2.0.21/include/event2/util.h Target /usr/local/include/event2/util.h already exists. You may need to delete it. To force the link and overwrite all other conflicting files, do: brew link --overwrite formula_name To list all files that would be deleted: brew link --overwrite --dry-run formula_name so I tried $ brew link --overwrite libevent but I got Linking /usr/local/Cellar/libevent/2.0.21... Warning: Could not link libevent. Unlinking... Error: Permission denied - /usr/local/include/event2/util.h so I tried $ sudo brew link --overwrite libevent Password: but I got Error: Cowardly refusing to sudo brew link You can use brew with sudo, but only if the brew executable is owned by root. However, this is both not recommended and completely unsupported so do so at your own risk. I'm stuck.

I'm trying to install an experimental version of [IguanaTex](https://github.com/ray851107/IguanaTexMac) on my Mac. I'm stuck and need some help. Here are the directions: 1. *install AppleScript (IguanaTex.scpt)* mkdir -p ~/Library/Application\ Scripts/com.microsoft.Powerpoint cp ./IguanaTex.scpt ~/Library/Application\ Scripts/com.microsoft.Powerpoint/IguanaTex.scpt This worked fine for me. 2. install IguanaTexHelper (libIguanaTexHelper.dylib) sudo mkdir -p '/Library/Application Support/Microsoft/Office365/User Content.localized/Add-Ins.localized' sudo cp ./libIguanaTexHelper.dylib '/Library/Application Support/Microsoft/Office365/User Content.localized/Add-Ins.localized/libIguanaTexHelper.dylib' **This is where I'm having an issue.** . I can't seem to find the file "libIguanaTexHelper.dylib" anywhere, and I can't find anything online about it. I've already made sure that my computer is showing hidden files. Can someone please help me clear this up?

I've just found that brew doctor gives me the following error : Please note that these warnings are just used to help the Homebrew maintainers with debugging if you file an issue. If everything you use Homebrew for is working fine: please don't worry and just ignore them. Thanks! Warning: Unbrewed dylibs were found in /usr/local/lib. If you didn't put them there on purpose they could cause problems when building Homebrew formulae, and may need to be deleted. Unexpected dylibs: /usr/local/lib/LibSideSyncOSX9.dylib /usr/local/lib/ss_conn_lib.dylib I am suspected that these are remnants of Samsung Sidesync program that I tried to use recently but it never worked, so I uninstalled it (using iTrash). Should I manually remove those two libraries or not(one of them is 1.7mb)? Thank you.

I have some custom modules that are generated using SWIG that are installed in a a local path eg: catkin_ws/devel/lib/python2.7/site-packages When I try to run a python script that uses one of these modules, I get an import error: "/usr/local/Cellar/python/2.7.13/Frameworks/Python.framework/Versions/2.7/lib/python2.7/importlib/__init__.py", line 37, in import_module __import__(name) ImportError: No module named _mag_manip I've figured out that this is due to the shebang at the top of the python script: #!/usr/bin/env python Because /usr/bin/env is protected by System Integrity Protection (SIP), the DYLD_LIBRARY_PATH environment variable is not loaded and some modules cannot be found. A fix is to replace the shebang with the hardcoded path of python ie: #!/usr/local/bin/python Since my code runs on Linux and Mac OSX, replacing the normal python shebang with a hardcoded path on mac is not a really good solution. Does anybody know what the best solution is here?

Let me start by saying I'm not traditionally a Mac user. I'm still trying to transition from Gentoo, where compiling everything from source was the default state, and the system wasn't actively trying to prevent me from doing "dangerous" things. I have a new machine running 10.11. I've gotten sqlplus 11.2.0.4 running, and am able to connect to a remote Oracle server with it. I now want to use DBD::Oracle to connect in a Perl script. I was able to get DBD::Oracle to compile/install by manually passing a version to Makefile.pl rather than using CPAN (or, more specifically, cpanm): perl Makefile.pl -V 11.2 make make install However, when I run the script using the default Perl (5.18.2 located at /usr/bin/perl), I get an error: ERROR OCIEnvNlsCreate. Check ORACLE_HOME (Linux) env var or PATH (Windows) and or NLS settings, permissions, etc. In case the issue lies with the system perl, I installed perlbrew and then installed 5.22.0 into ~/perl5. Using the perlbrew Perl, I get a different error: dyld: lazy symbol binding failed: Symbol not found: _OCIAttrSet Referenced from: /Users/jrittenh/perl5/perlbrew/perls/perl-5.22.0/lib/site_perl/5.22.0/darwin-2level/auto/DBD/Oracle/Oracle.bundle Expected in: dynamic lookup dyld: Symbol not found: _OCIAttrSet Referenced from: /Users/jrittenh/perl5/perlbrew/perls/perl-5.22.0/lib/site_perl/5.22.0/darwin-2level/auto/DBD/Oracle/Oracle.bundle Expected in: dynamic lookup Trace/BPT trap: 5 I've tried: * both 32-bit and 64-bit Oracle libraries * symlinking libraries from the specific version (libclntsh.dylib.11.1 => libclntsh.dylib) * fixing hard-coded paths to /ade * creating a 'mesg' folder and made it writeable in $ORACLE_HOME * setting ORACLE_HOME, LD_LIBRARY_PATH, DYLD_LIBRARY_PATH, and PATH to include /usr/local/oracle_client from /etc/profile, /etc/bashrc/, ~/.bashrc, and ~/.bash_profile Is there something else I can try? What am I missing?

I am using iTunes on Windows 10. There are multiple "copies" (air quotes) of most (not all) audiobooks in iTunes. I used air quotes because when I search the PC for the books, there are only single instances. Another post (https://apple.stackexchange.com/questions/64097/why-does-itunes-create-multiple-exact-copies-of-my-music-files-and-hide-them-fro) asked a similar question about music files. The answer for that post alluded to a playlist problem. But (a) I do NOT have any playlists for my audiobooks, and (b) that post is old, dating to Sept. 2012. I am running version 12.7.2. When I select a copy of a book and drag it to my iphone, I can see that the copies are numbered 1, 2, 3, ... When I try to delete any one of those copies, then all copies are deleted. What can I do clean up the lists so that I only see one copy of the book?

In shell scripts used for unit testing with dynamic libraries in a directory other than the typical @rpath, I have previously been able to set DYLD_FALLBACK_LIBRARY_PATH to set the directory containing the libraries. Under 10.11.1, bash seems to ignore attempts to set this environment variable: $ sh -x testscript.sh + DYLD_FALLBACK_LIBRARY_PATH=/Users/something/testinglibs + export DYLD_FALLBACK_LIBRARY_PATH + exec printenv and DYLD_FALLBACK_LIBRARY_PATH is not present in printenv's output. Is this a security-related hack in 10.11's shell? I haven't been able to find this change documented in man pages or online.

I have macports 2.5.2 already installed. I have then installed Gnucash 3.2 dmg from gnucash.org. When I run gnucash, I have the error: > objc: Class GNotificationCenterDelegate is implemented in both > /Applications/Gnucash.app/Contents/Resources/lib/libgio-2.0.0.dylib > (0x10c333670) and /opt/local/lib/libgio-2.0.0.dylib (0x117a41620). One > of the two will be used. Which one is undefined. Segmentation fault: > 11 /opt/local/lib/libgio-2.0.0.dylib belongs to glib2 macports package. **If I rename /opt/local/lib/libgio-2.0.0.dylib, gnucash works fine**. But this can't be a solution. I've also tried to set DYLB_LIBRARY_PATH to > /Applications/Gnucash.app/Contents/Resources/lib But this way I have the error > Segmentation fault: 11 Is there a way to run gnucash while macports installed ?

I have a Fortran program which I run on my macOS 10.12.6. The code is compiled using a Makefile, in which all the libraries included have their path included as well. For one of the libraries (and only one) my executable can't seem to be able to find it. My code compiles correctly but I get this runtime error at the beginning of the program: dyld: Library not loaded: libslatec.so Referenced from: path_to_PWD/exe/executable Reason: image not found Here's how I am adding my library: LIB_SLATEC = -L$(HOME)/usr/lib/slatec/lib -lslatec If I do ls $HOME/usr/lib/slatec/lib I get: libslatec.so. Running otool -L path_to_exe | grep slatec I get: libslatec.so (compatibility version 0.0.0, current version 0.0.0). I see that there is no path specified there. If I copy the library into my $PWD the code runs as expected. I have also tried adding the library path to LD_LIBRARY_PATH but that did not change anything. Any idea what the problem is?

I'm on macOS Sierra 10.12.5, and trying to add a library to the usr/lib folder (for an OpenFrameworks project I'm working on). I've seen similar questions here , here , and here . However, none of the suggested answers seem to have any effect. On Sierra, it seems that when rebooting in recovery mode, there is no longer any way to access a Terminal window (unless I'm mistaken), so System Integrity Protection can't be disabled by running csrutil disable. Does anyone have any suggestions?