fetchmail-devel

[fetchmail-devel] [PATCHes] Misc. bug fixes

[Miloslav T. <mi...@re...>]

Hello,
The warning cleanups have allowed me to notice bugs on some more exotic
architectures which were previously lost in the noise:

- in several places the *printf format string doesn't agree with the
  parameter types
- glibc doesn't export res_* on architectures with newer ABI, only
  __res_* and defines macros in <resolv.h>, thus configure would not
  detect res_* presence.

The attached patches fix these problems.
Thanks,
	Mirek

Re: [fetchmail-devel] [PATCHes] Misc. bug fixes

[Matthias A. <mat...@gm...>]

Miloslav Trmac <mi...@re...> writes:

> Hello,
> The warning cleanups have allowed me to notice bugs on some more exotic
> architectures which were previously lost in the noise:
>
> - in several places the *printf format string doesn't agree with the
>   parameter types
> - glibc doesn't export res_* on architectures with newer ABI, only
>   __res_* and defines macros in <resolv.h>, thus configure would not
>   detect res_* presence.
>
> The attached patches fix these problems.

Merged for 6.3.4, thanks.

-- 
Matthias Andree

Re: [fetchmail-devel] [PATCHes] Misc. bug fixes

[Matthias A. <mat...@gm...>]

Miloslav Trmac <mi...@re...> writes:

> +old_libs=$LIBS

This should be old_LIBS="$LIBS" with capitalized LIBS part in the
variable name, in order to match...

> +for lib in '' -lresolv; do
> +    if test -z "$lib"; then
> +       AC_MSG_CHECKING([for res_search in libc])
> +    else
> +       AC_MSG_CHECKING([for res_search in $lib])
> +    fi
> +    LIBS="$old_LIBS $lib"

...this part      ^^^^

Else we'd lose whatever was in LIBS at the time (for instance -lsocket
-lnsl on Solaris, where I found the problem).

-- 
Matthias Andree