[PATCH 0/2] incompatible-poiner-types bug fixes

[PATCH 0/2] incompatible-poiner-types bug fixes

[Nicholas Vinson]

This patch series fixes several bugs found by compiling with
-Werror=incompatible-pointer-types. This setting will be gcc's default
behavior starting with gcc-14.

See:
https://wiki.gentoo.org/wiki/Modern_C_porting#What_changed.3F
https://inbox.sourceware.org/gcc-patches/cover.1700473918.git.fweimer@redhat.com/
for details.

Nicholas Vinson (2):
  Fix incompatible-pointer-types bugs
  Fix incompatible-pointer-types configure test bugs

 Src/Modules/termcap.c | 36 ++++++++++++++++++++----------------
 configure.ac          | 12 ++++++------
 2 files changed, 26 insertions(+), 22 deletions(-)

-- 
2.43.0

Re: [PATCH 0/2] incompatible-poiner-types bug fixes

[Bart Schaefer]

On Sun, Dec 10, 2023 at 1:50 PM Nicholas Vinson <nvinson234@gmail.com> wrote:
>
> This patch series fixes several bugs found by compiling with
> -Werror=incompatible-pointer-types.

We've already got a patch for this from Florian Weimer
<fweimer@redhat.com> in workers/52383.  Is the zboolcodes naming and
BOOLCODES define in Src/Modules/termcap.c actually necessary?

Re: [PATCH 0/2] incompatible-poiner-types bug fixes

[Nicholas Vinson]

If configure incorrectly decides (or is incorrectly told) that boolcodes 
or numcodes do not exist, you end up with a naming conflict at 
compile-time. This happened while I was testing my changes. I suspect, 
but don't know for sure, that this also happened at some point with 
strcodes since the handling of that variable is different than how 
boolcodes and numcodes are handled. Regardless, the naming change 
prevents the naming conflict, and the defines prevent added complexity 
on the for-loops and simplifies the strcodes for-loop handling.

Moreover, these changes align the handling of all three variables, 
boolcodes, numcodes, strcodes, in a type-safe and name-safe manner.

Thanks,

Nicholas Vinson

On 12/10/23 17:45, Bart Schaefer wrote:
> On Sun, Dec 10, 2023 at 1:50 PM Nicholas Vinson <nvinson234@gmail.com> wrote:
>> This patch series fixes several bugs found by compiling with
>> -Werror=incompatible-pointer-types.
> We've already got a patch for this from Florian Weimer
> <fweimer@redhat.com> in workers/52383.  Is the zboolcodes naming and
> BOOLCODES define in Src/Modules/termcap.c actually necessary?

Re: [PATCH 0/2] incompatible-poiner-types bug fixes

[Oliver Kiddle]

Nicholas Vinson wrote:
> If configure incorrectly decides (or is incorrectly told) that boolcodes 
> or numcodes do not exist, you end up with a naming conflict at 
> compile-time. This happened while I was testing my changes. I suspect, 

It may be for the best that we get errors if configure has got things
wrong.

> Moreover, these changes align the handling of all three variables, 
> boolcodes, numcodes, strcodes, in a type-safe and name-safe manner.

If adding const makes the code align with ncurses then that could be
useful. But the #defines seem ugly and unnecessary.

Oliver