failed to build in gentoo

[rpolasek]

hi, trying to build the latest version from git in gentoo...

FAILED: CMakeFiles/c3c.dir/src/compiler/sema_stmts.c.o /usr/bin/x86_64-pc-linux-gnu-gcc -DCURL_FOUND=1 -DLLVM_AVAILABLE=1 -DTB_AVAILABLE=0 -I/usr/lib/llvm/19/include -I/var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src -I/var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999_build -I/var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/wrapper/include -I/var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/dependencies/miniz -march=native -O2 -pipe -std=gnu11 -D_GNU_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -pthread -Wall -Werror -Wno-unknown-pragmas -Wno-unused-result -Wno-unused-function -Wno-unused-variable -Wno-unused-parameter -MD -MT CMakeFiles/c3c.dir/src/compiler/sema_stmts.c.o -MF CMakeFiles/c3c.dir/src/compiler/sema_stmts.c.o.d -o CMakeFiles/c3c.dir/src/compiler/sema_stmts.c.o -c /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/compiler/sema_stmts.c In function ‘sema_analyse_ct_foreach_stmt’, inlined from ‘sema_analyse_statement_inner’ at /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/compiler/sema_stmts.c:3139:11: /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/compiler/sema_stmts.c:2805:32: error: ‘count’ may be used uninitialized [-Werror=maybe-uninitialized] 2805 | for (unsigned i = 0; i < count; i++) | ~~^~~~~~~ /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/compiler/sema_stmts.c: In function ‘sema_analyse_statement_inner’: /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/compiler/sema_stmts.c:2727:18: note: ‘count’ was declared here 2727 | unsigned count; | ^~~~~ cc1: all warnings being treated as errors

[lerno]

That looks like incorrect diagnostics because it fails to properly analyse the flow. But I changed it so hopefully it now compiles for you.

[rpolasek]

@lerno, thank you for the prompt fix. unfortunately, there is next problem with compilation but i can feel we're close and close to the final success build in gentoo. (btw. i can provide you c3-9999.ebuild for gentoo at the end).

here is a piece of log message with an error but i'm enclosing a full build log for better investigaton...

FAILED: CMakeFiles/c3c.dir/src/utils/file_utils.c.o /usr/bin/x86_64-pc-linux-gnu-gcc -DCURL_FOUND=1 -DLLVM_AVAILABLE=1 -DTB_AVAILABLE=0 -I/usr/lib/llvm/19/include -I/var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src -I/var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999_build -I/var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/wrapper/include -I/var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/dependencies/miniz -march=native -O2 -pipe -std=gnu11 -D_GNU_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -D__STDC_LIMIT_MACROS -pthread -Wall -Werror -Wno-unknown-pragmas -Wno-unused-result -Wno-unused-function -Wno-unused-variable -Wno-unused-parameter -MD -MT CMakeFiles/c3c.dir/src/utils/file_utils.c.o -MF CMakeFiles/c3c.dir/src/utils/file_utils.c.o.d -o CMakeFiles/c3c.dir/src/utils/file_utils.c.o -c /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/file_utils.c /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/file_utils.c: In function ‘file_append_path_temp’: /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/file_utils.c:509:53: error: ‘%s’ directive output may be truncated writing up to 16382 bytes into a region of size between 1 and 16382 [-Werror=format-truncation=] 509 | snprintf(path_buffer, PATH_BUFFER_SIZE, "%s/%s", path, name); | ^~ In file included from /usr/include/stdio.h:970, from /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/common.h:9, from /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/file_utils.c:7: In function ‘snprintf’, inlined from ‘file_append_path_temp’ at /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/file_utils.c:509:2: /usr/include/bits/stdio2.h:68:10: note: ‘__builtin___snprintf_chk’ output between 3 and 32766 bytes into a destination of size 16384 68 | return __builtin___snprintf_chk (__s, __n, __USE_FORTIFY_LEVEL - 1, | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 69 | __glibc_objsize (__s), __fmt, | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 70 | __va_arg_pack ()); | ~~~~~~~~~~~~~~~~~ /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/file_utils.c: In function ‘file_append_path_temp’: /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/file_utils.c:514:52: error: ‘%s’ directive output may be truncated writing up to 16381 bytes into a region of size between 2 and 16383 [-Werror=format-truncation=] 514 | snprintf(path_buffer, PATH_BUFFER_SIZE, "%s%s", path, name); | ^~ In function ‘snprintf’, inlined from ‘file_append_path_temp’ at /var/tmpfs/portage/dev-lang/c3-9999/work/c3-9999/src/utils/file_utils.c:514:2: /usr/include/bits/stdio2.h:68:10: note: ‘__builtin___snprintf_chk’ output between 2 and 32764 bytes into a destination of size 16384 68 | return __builtin___snprintf_chk (__s, __n, __USE_FORTIFY_LEVEL - 1, | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 69 | __glibc_objsize (__s), __fmt, | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 70 | __va_arg_pack ()); | ~~~~~~~~~~~~~~~~~ cc1: all warnings being treated as errors

[rpolasek]

i've just investigated the code and you don't handle a buffer overflow well at line 500 in the file file_utils.c. the check should be path_len + name_len + 2 >= PATH_BUFFER_SIZE (extra 1 for the / or \, and 1 for \0). also path_buffer[name_len + path_len + 1] = 0; is redundant since snprintf ensures a null-terminated string. long story short - i've rewritten whole function file_append_path_temp and now i'm able to compile it in gentoo.

const char *file_append_path_temp(const char *path, const char *name) {
    size_t path_len = strlen(path);
    if (path_len == 0) return name; // if no path, return name directly

    size_t name_len = strlen(name);

    // ensure there's space for path, separator, name, and null terminator
    if (path_len + name_len + 2 > PATH_BUFFER_SIZE) {
        error_exit("Error generating path from %s and %s: buffer max size exceeded.", path, name);
        return NULL;
    }

    // determine if we need a separator
    char separator = '/';
#if defined(_WIN32)
    separator = '\\';
#endif

    // format the string safely
    int written;
    if (path[path_len - 1] == '/' || path[path_len - 1] == '\\') {
        written = snprintf(path_buffer, PATH_BUFFER_SIZE, "%s%s", path, name);
    } else {
        written = snprintf(path_buffer, PATH_BUFFER_SIZE, "%s%c%s", path, separator, name);
    }

    // check if truncation occurred
    if (written < 0 || (size_t)written >= PATH_BUFFER_SIZE) {
        error_exit("Error generating path from %s and %s: snprintf truncation occurred.", path, name);
        return NULL;
    }

    return path_buffer;
}

[rpolasek]

as soon as the code above is fixed you can provide all gentoo users c3-9999.ebuild and c3-0.6.6.ebuild.

[lerno]

I modified it slightly, hopefully it still passes.

[rpolasek]

i've updated the links for my ebuilds above and i can confirm it works great in gentoo now. however, i have to ask you if your changed code below is really ok primarily for non windows platform (separator = '/'):

bool insert_separator = path[path_len - 1] == '/' || path[path_len - 1] == separator;

[lerno]

Yes, should be fine since both comparisons are the same.