Re: [lmi] [lmi-commits] master bdeeca5c 5/5: Improve technique to display disabled warnings

[Greg Chicares]

On 6/12/22 17:32, Vadim Zeitlin wrote:
[...]
> a link to my small project which might be handy:
> 
>       https://github.com/vadz/gcc-warnings-tools
> 
> and direct links to the tables generated by it:
> 
>       https://github.com/vadz/gcc-warnings-tools/wiki/gcc-warnings
>       https://github.com/vadz/gcc-warnings-tools/wiki/all-gcc-warnings
[...]
> warnings added in a new gcc release. E.g., after just updating the pages
> above for gcc 12, looking at the changes you can see that it added the
> following new warnings not enabled by -Wall (or by default): array-compare,
> infinite-recursion, openacc-parallelism, trivial-auto-var-init and
> use-after-free, some of which might be useful to explore when lmi starts
> using it.

I've explicitly enabled these two:
  openacc-parallelism
  trivial-auto-var-init
and they engender no new diagnostic anywhere in lmi.

The other three:
  array-compare
  infinite-recursion
  use-after-free
do seem to be enabled by '-Wall', according to the current gcc
documentation (which may well have changed since last summer).

The use-after-free one is weird: the "summary" page
  https://gcc.gnu.org/onlinedocs/gcc/Option-Summary.html#Option-Summary
speaks only of
  -Wno-analyzer-use-after-free
whereas the "warnings" page
  https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
says that
  -Wuse-after-free=2
is enabled by '-Wall'. At first, I thought that the "analyzer-"
part might have been a typographical error, but the "warnings"
page clearly distinguishes '-Wanalyzer-infinite-recursion' from
'-Winfinite-recursion', so maybe the two use-after-free options
are likewise distinct, and the shorter '-Wuse-after-free' one
defectively lacks any explanation (e.g., of what "=2" means).