‘surrogates_to_codepoint’ warnings

‘surrogates_to_codepoint’ warnings

[martin rudalics]

Building master currently gets me lots of

../../src/coding.h:680:1: warning: no previous prototype for ‘surrogates_to_codepoint’ [-Wmissing-prototypes]

Ignore them?

martin

Re: ‘surrogates_to_codepoint’ warnings

[Eli Zaretskii]

> Date: Mon, 15 Jan 2018 10:39:59 +0100
> From: martin rudalics <rudalics@gmx.at>
> 
> Building master currently gets me lots of
> 
> ../../src/coding.h:680:1: warning: no previous prototype for ‘surrogates_to_codepoint’ [-Wmissing-prototypes]
> 
> Ignore them?

No, never.  Should be fixed now.

Thanks.

Re: ‘surrogates_to_codepoint’ warnings

[Richard Copley]

On 15 January 2018 at 12:16, Eli Zaretskii <eliz@gnu.org> wrote:
>> Date: Mon, 15 Jan 2018 10:39:59 +0100
>> From: martin rudalics <rudalics@gmx.at>
>>
>> Building master currently gets me lots of
>>
>> ../../src/coding.h:680:1: warning: no previous prototype for ‘surrogates_to_codepoint’ [-Wmissing-prototypes]
>>
>> Ignore them?
>
> No, never.

:)

>  Should be fixed now.

Below are the C warnings I currently get when building the master
branch, on MSYS2. I've included the warnings from indent.c below
because their text is quite different. I don't know whether they're
any more useful than their previously-discussed emacs-26 counterparts.
The emacs-26 branch built cleanly last time I checked, except for the
warnings in indent.c discussed here:
<https://lists.gnu.org/archive/html/emacs-devel/2017-10/msg00149.html>

  CC       indent.o
In file included from indent.c:31:0:
indent.c: In function 'scan_for_column':
disptab.h:43:7: warning: potential null pointer dereference [-Wnull-dereference]
    ? (NILP ((dp)->ascii)    \
       ^~~~~~~~~~~~~~~~~~
indent.c:297:40: note: in expansion of macro 'DISP_CHAR_VECTOR'
    width = sanitize_char_width (ASIZE (DISP_CHAR_VECTOR (dp, ch))); \
                                        ^~~~~~~~~~~~~~~~
indent.c:516:8: note: in expansion of macro 'MULTIBYTE_BYTES_WIDTH'
        MULTIBYTE_BYTES_WIDTH (p, buffer_display_table (), b, wd);
        ^~~~~~~~~~~~~~~~~~~~~
disptab.h:43:7: warning: potential null pointer dereference [-Wnull-dereference]
    ? (NILP ((dp)->ascii)    \
       ^~~~~~~~~~~~~~~~~~
indent.c:297:40: note: in expansion of macro 'DISP_CHAR_VECTOR'
    width = sanitize_char_width (ASIZE (DISP_CHAR_VECTOR (dp, ch))); \
                                        ^~~~~~~~~~~~~~~~
indent.c:516:8: note: in expansion of macro 'MULTIBYTE_BYTES_WIDTH'
        MULTIBYTE_BYTES_WIDTH (p, buffer_display_table (), b, wd);
        ^~~~~~~~~~~~~~~~~~~~~
disptab.h:43:7: warning: potential null pointer dereference [-Wnull-dereference]
    ? (NILP ((dp)->ascii)    \
       ^~~~~~~~~~~~~~~~~~
indent.c:297:40: note: in expansion of macro 'DISP_CHAR_VECTOR'
    width = sanitize_char_width (ASIZE (DISP_CHAR_VECTOR (dp, ch))); \
                                        ^~~~~~~~~~~~~~~~
indent.c:516:8: note: in expansion of macro 'MULTIBYTE_BYTES_WIDTH'
        MULTIBYTE_BYTES_WIDTH (p, buffer_display_table (), b, wd);
        ^~~~~~~~~~~~~~~~~~~~~
disptab.h:43:7: warning: potential null pointer dereference [-Wnull-dereference]
    ? (NILP ((dp)->ascii)    \
       ^~~~~~~~~~~~~~~~~~
indent.c:297:40: note: in expansion of macro 'DISP_CHAR_VECTOR'
    width = sanitize_char_width (ASIZE (DISP_CHAR_VECTOR (dp, ch))); \
                                        ^~~~~~~~~~~~~~~~
indent.c:516:8: note: in expansion of macro 'MULTIBYTE_BYTES_WIDTH'
        MULTIBYTE_BYTES_WIDTH (p, buffer_display_table (), b, wd);
        ^~~~~~~~~~~~~~~~~~~~~

CC       intervals.o
intervals.c: In function 'offset_intervals':
intervals.c:878:20: warning: potential null pointer dereference
[-Wnull-dereference]
   if (position == i->position || eobp)
                   ~^~~~~~~~~~
make[1]: Leaving directory '/c/projects/emacs/src'
make[1]: Entering directory '/c/projects/emacs/src'
  CC       process.o
process.c: In function 'connect_network_socket':
process.c:208:10: warning: 'family' may be used uninitialized in this
function [-Wmaybe-uninitialized]
   return close_on_exec (socket (domain, type, protocol));
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
process.c:3320:7: note: 'family' was declared here
   int family;
       ^~~~~~

Re: ‘surrogates_to_codepoint’ warnings

[Paul Eggert]

Richard Copley wrote:
> Below are the C warnings I currently get when building the master
> branch, on MSYS2.

These appear to be false alarms. Which compiler and version are you using, and 
how are you invoking 'configure'? I ask because often we don't bother to pacify 
false alarms from older or less-used compilers, or compilers invoked with 
unusual options.

Re: ‘surrogates_to_codepoint’ warnings

[Richard Copley]

On 15 January 2018 at 19:30, Paul Eggert <eggert@cs.ucla.edu> wrote:
> Richard Copley wrote:
>>
>> Below are the C warnings I currently get when building the master
>> branch, on MSYS2.
>
>
> These appear to be false alarms. Which compiler and version are you using,
> and how are you invoking 'configure'? I ask because often we don't bother to
> pacify false alarms from older or less-used compilers, or compilers invoked
> with unusual options.

Same as last time you asked :)

GCC 7.2.

./configure --config-cache --with-modules --without-pop 'CFLAGS=-O3 -ggdb3'

Re: ‘surrogates_to_codepoint’ warnings

[Paul Eggert]

Richard Copley wrote:
> GCC 7.2.
> 
> ./configure --config-cache --with-modules --without-pop 'CFLAGS=-O3 -ggdb3'

I tried that on Fedora 27, which has GCC 7.2.1 20170915 (Red Hat 7.2.1-2), and 
got some of the warnings that you mention. I suggest using -O2 instead, as this 
is the default optimization level and it made the warnings go away for me. I'd 
rather not contort the code to pacify false alarms for every possible 
combination of GCC options, as it's too much work for too little benefit.

Re: ‘surrogates_to_codepoint’ warnings

[Richard Copley]

On 15 January 2018 at 21:38, Paul Eggert <eggert@cs.ucla.edu> wrote:
> Richard Copley wrote:
>>
>> GCC 7.2.
>>
>> ./configure --config-cache --with-modules --without-pop 'CFLAGS=-O3
>> -ggdb3'
>
>
> I tried that on Fedora 27, which has GCC 7.2.1 20170915 (Red Hat 7.2.1-2),
> and got some of the warnings that you mention. I suggest using -O2 instead,
> as this is the default optimization level and it made the warnings go away
> for me. I'd rather not contort the code to pacify false alarms for every
> possible combination of GCC options, as it's too much work for too little
> benefit.

OK, thanks. I get no warnings at -O2.

Re: ‘surrogates_to_codepoint’ warnings

[Eli Zaretskii]

> From: Richard Copley <rcopley@gmail.com>
> Date: Mon, 15 Jan 2018 22:19:07 +0000
> Cc: martin rudalics <rudalics@gmx.at>, Eli Zaretskii <eliz@gnu.org>,
> 	Emacs Development <emacs-devel@gnu.org>
> 
> OK, thanks. I get no warnings at -O2.

One more reason not to use -O3, I guess.  IME, that option is not for
large and complex interactive programs such as Emacs.

Re: ‘surrogates_to_codepoint’ warnings

[martin rudalics]

> Should be fixed now.

Thank you, martin