Arch Linux User Repository

For ease of upstreaming, here are two patches made from the guidance in comments below. quickbms builds for me with these. I am using GCC.

$ cat 0003-disable-atomic.patch

--- src/extra/stdatomic.h.orig
+++ src/extra/stdatomic.h
@@ -67,7 +67,7 @@
 #endif

 #if !defined(__CLANG_ATOMICS) && !defined(__GNUC_ATOMICS)
-#if __has_feature(c_atomic)
+#if __has_feature(nonexistent_feature)
 #define    __CLANG_ATOMICS
 #elif __GNUC_PREREQ__(4, 7)
 #define    __GNUC_ATOMICS

$ cat 0004-disable-werror-failures.patch

--- src/Makefile.orig       2025-11-02 19:57:29.283380514 -0800
+++ src/Makefile    2025-11-02 19:58:44.116358943 -0800
@@ -1,6 +1,7 @@
 EXE                = quickbms
 # -m32 because QuickBMS has been tested only on 32bit systems and gives problems using 64bit native code
 CFLAGS     += -m32 -s -O2 -fstack-protector-all -fno-unit-at-a-time -fno-omit-frame-pointer -fno-pie -fPIC -w
+CFLAGS  += -Wno-error=incompatible-pointer-types -Wno-error=implicit-function-declaration
 # Add -DQUICKBMS64 to CDEFS for compiling quickbms_4gb_files
 # For static builds on Linux (no MacOSX due to -lcrt0 error) add: -static
 # -ldl is in CDEFS to avoid undefined references.

Edit: pasting these is terrible and is probably going to break on whitespace, so I'll get a git up with the diffs to be clearer. Patches are visible as commits here: https://github.com/Canar/quickbms/commits/master/

Numerous compiler errors can be suppressed with CFLAGS="-Wno-error=incompatible-pointer-types -Wno-error=implicit-function-declaration".

Linker errors related to __c11_atomic_load and __c11_atomic_store can be fixed with something like sed -i 's/c_atomic/nonexistent_feature/' extra/stdatomic.h. This BSD header erroneously detects clang because __has_feature(c_atomic) is true.

This package doesn't build with clang. Please, use GCC explicitly:

build() {
  cd src

  CC=GCC make
}

You could mention game modders in the description as well, as target audience. Their use case is the prime motivation behind this project.

RSA_SSLV23_PADDING was deprecated due to CVE-2021-23839. Here is a patch that fixes it (Seems consistent with how other software is doing it) 0001-Fix-OpenSSL-V21-padding-deprecation.patch

error: ‘RSA_SSLV23_PADDING’ undeclared (first use in this function); did you mean ‘RSA_NO_PADDING’

I bypassed this error by modifying the Makefile. Removed USE_OPENSSL = 1 and it compiles, but some features are probably disabled by this and I'm not sure about which are.

Builds are failing for me with this

error: ‘RSA_SSLV23_PADDING’ undeclared (first use in this function); did you mean ‘RSA_NO_PADDING’

Looks like a change made in openssl 3? I do have openssl-1.1 installed as well but still no good, maybe I'm doing something wrong.

while just want to notify 0.11.0 is out. http://aluigi.zenhax.com/papers/quickbms-src-0.11.0.zip. also, thanks for the package, it looks like not easy to maintenance a package.

Yes and no. I've looked at the patch you provided but it didn't seem like a correct solution because I think it will definitely break functionality within that file. Of course you can use this presumably incorrect version as long as you don't use the exact functionality, but I don't want to provide everyone with this version. And I'm a bit tired of fixing quickbms with every release because upstream breaks in completely weird ways all the time (and the ridiculous makefile with compiles everyone in one go also doesn't help).

is this package still being maintained? (i don't intend this to sound rude or mean)