Jim Meyering wrote: > Thank you for fixing those things for both gzip and diffutils!
It is at the limit of what kind of constraints packages may impose on gnulib. If other kinds of warnings appear, which don't lead to code improvements in gnulib, I'll kindly ask the maintainer of such packages to remove some warning options from the set of warnings for which a -Werror is requested. More generally, I think it would be useful to replace the line gl_WARN_ADD([-Werror], [WERROR_CFLAGS]) found in gzip/configure.ac and diffutils/configure.ac with options that request an error only for specific compiler warnings. Recall that the 'manywarnings' module is built in such a way that when a new GCC release is made, all its new warnings are enabled. Paul then updates the 'manywarnings' module, by disabling those warning options that clearly produce more false positives than valuable reports. Thus some new warnings will typically appear: good ones as well as silly ones (like the "format not a string literal and no format arguments" warning). And these new warnings then cause "make distcheck" to fail. While I understand a maintainer's desire to have warning-free compilation on their own machine, I also think it is not good if "make distcheck" starts to break for people who use a newer compiler than the maintainer. Therefore, how about applying -Werror only to a subset of all warnings, hand-picked by the maintainer? Bruno