[Bug c/60296] Confusing -Wformat warning on invalid format string

[manu at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60296

Manuel López-Ibáñez <manu at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |manu at gcc dot gnu.org
         Resolution|---                         |INVALID

--- Comment #2 from Manuel López-Ibáñez <manu at gcc dot gnu.org> ---
To be honest, I find the messages by gcc far clearer (except for missing
precise locations). In particular, if we think about suggesting a fix, gcc
could suggest to replace %. with %.d to match '9' and then the second warning
will still be valid (since there are no arguments to match %*d). Currently, gcc
tries to match 9 to %. and skips both on error, that is why it complains about
a missing argument for %d. GCC could not skip the argument '9' and try to match
%*d with 9. I can imagine cases where one strategy is clearer than the other
and viceversa. However, the clang warning does not actually say what is wrong
at all. It does not seem very helpful.

About your questions, yes %d refers to %*d in the format string, and the %d
printed by the warning is not read from the string but hard-coded in the gcc
sources.

What printf does when given invalid arguments is not up to GCC but the C
library that implements printf.

I will close this as invalid then. Improving the location info is what will
make these cases far clearer (PR52952). Then, we can print:

warning: conversion lacks type at end of format [-Wformat=]
   printf("%.%*d\n", 9);
           ~^
note: use '%.d' to match argument of type 'int'
   printf("%.%*d\n", 9);
           ~^        ^
           %.d

warning: format ‘%d’ expects a matching ‘int’ argument [-Wformat=]
   printf("%.%*d\n", 9);
               ^