Remove indirection when calling dbuf_printf

[saghul]

No description provided.

[chqrlie]

The reason for this indirection is to prevent compiler warnings about the format string not being a string constant.

[saghul]

What compiler version triggers it? I haven't seen them in the CI run!

[saghul]

How is that board meeting going? 😅

[chqrlie]

How is that board meeting going? 😅

That was a rocky one...

It is very surprising to me that you were able to remove the indirection: it means the compiler(s) do not emit a warning for code that invokes dbuf_printf with a format that is not a string literal. This is fishy! Did you explicitly disable this warning or did the recent change on the printf_like attribute disable the format string verifications inadvertently?

[saghul]

I guess you mean -Wformat-nonliteral ? Looks like that is not enabled even with -Wall, one needs to set -Wformat=2 : https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html

Now, if I set that, even without this PR:

/Users/saghul/src/quickjs/libbf.c:3973:44: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
 3973 |                             dbuf_printf(s, fmt, (n - 1) * radix_bits);
      |                                            ^~~
/Users/saghul/src/quickjs/libbf.c:3979:44: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
 3979 |                             dbuf_printf(s, fmt,
      |                                            ^~~
2 errors generated.
make[4]: *** [CMakeFiles/qjs.dir/libbf.c.o] Error 1
make[4]: *** Waiting for unfinished jobs....
/Users/saghul/src/quickjs/quickjs.c:6865:33: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
 6865 |     vsnprintf(buf, sizeof(buf), fmt, ap);
      |                                 ^~~
/Applications/Xcode-16.0.0.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX15.0.sdk/usr/include/secure/_stdio.h:75:63: note: expanded from macro 'vsnprintf'
   75 |   __builtin___vsnprintf_chk (str, len, 0, __darwin_obsz(str), format, ap)
      |                                                               ^~~~~~
/Users/saghul/src/quickjs/quickjs.c:6940:35: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
 6940 |     return JS_ThrowTypeError(ctx, fmt,
      |                                   ^~~
/Users/saghul/src/quickjs/quickjs.c:6948:37: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
 6948 |     return JS_ThrowSyntaxError(ctx, fmt,
      |                                     ^~~
3 errors generated.

So it looks like we are alrady in that territory.

[gschwind]

Hello,

maybe some comment around the indirection could be useful such as:
keep it to avoid misleading printf format error

Best regards

[saghul]

This PR makes that error surface, yes:

/Users/saghul/src/quickjs/libbf.c:3973:44: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
 3973 |                             dbuf_printf(s, fmt, (n - 1) * radix_bits);
      |                                            ^~~
/Users/saghul/src/quickjs/libbf.c:3979:44: error: format string is not a string literal [-Werror,-Wformat-nonliteral]
 3979 |                             dbuf_printf(s, fmt,
      |                                            ^~~

My point is that we are not compiling with those flags anyway, and we already have the problem in other palces so we are not going to turn that flag on.

[saghul]

Closing in favor of #857