Re: [Tuxpaint-devel] Per-locale default font

[Bill K. <nb...@so...>]

On Mon, Jun 19, 2023 at 08:26:46AM +0900, Shin-ichi TOYAMA wrote:
> On Sun, 18 Jun 2023 13:38:35 -0700, Bill Kendrick wrote:
> >It's doubtful I'll get to it today, as I've got an engagement in a
> >bit, but it's nice to get use closer to 'zero warnings' when
> >compiling.
> >
> >At least, that's what it looks like for ME, just doing a plain
> >`make` here on my Kubuntu 22.04 system.  If anyone else is still
> >seeing compiler warnings, please share them so we can try and
> >clean up the codebase.  It's nice for _real_ errors to not be
> >buried in a sea of warnings. :)
> 
> Here are warnings on Windows so far.
> I will see them when I have time.
> 
> src/tuxpaint.c:234:2: warning: #warning "Attempting to define strcasestr(); if errors, build with -DHAVE_STRCASESTR" [-Wcpp]
>   234 | #warning "Attempting to define strcasestr(); if errors, build with -DHAVE_STRCASESTR"
>       |  ^~~~~~~

Oh, this is our own warning.  It's informative, and wow it's
from forever ago! :^D

  /* Check if features.h did its 'magic', in which case strcasestr() is
     likely available; if not using GNU, you can set HAVE_STRCASESTR to
     avoid trying to redefine it -bjk 2006.06.02 */

  #if !defined(__USE_GNU) && !defined(HAVE_STRCASESTR)
  #warning "Attempting to define strcasestr(); if errors, build with -DHAVE_STRCASESTR"

  ... then we declare our own strcasestr() func ...

I, uh, don't see a "features.h" in 'tuxpaint' source these days.
I'm not sure the best solution here, but I'll ignore it for now. :)


> src/tuxpaint.c: In function 'mainloop':
> src/tuxpaint.c:3747:21: warning: the comparison will always evaluate as 'true' for the address of 'sizes' will never be NULL [-Waddress]
>  3747 |                 if (magics[magic_group][cur_thing].sizes)
>       |                     ^~~~~~
> src/tuxpaint.c:1581:7: note: 'sizes' declared here
>  1581 |   int sizes[MAX_MODES];         /* Whether magic tool accepts sizes */
>       |       ^~~~~
<snip repeats of similar warnigns>

Ah whoops wow yes, these are places where I forgot to change the
reference from "sizes" to "sizes[...]", when I turned it into an
array.  My 'gcc' shows no complaints, unless I actually make it
explicitly test whether it == 1, in which case I get

  warning: comparison between pointer and integer

Just committed an attempt to mend this. :whew:


> src/tuxpaint.c: In function '_load_svg':
> src/tuxpaint.c:21025:5: warning: 'rsvg_handle_close' is deprecated: Use 'rsvg_handle_read_stream_sync' instead [-Wdeprecated-declarations]
> 21025 |     rsvg_handle_close(rsvg_handle, &gerr);
>       |     ^~~~~~~~~~~~~~~~~
> In file included from src/tuxpaint.c:517:
> C:/msys64/mingw64/include/librsvg-2.0/librsvg/rsvg.h:605:10: note: declared here
>   605 | gboolean rsvg_handle_close (RsvgHandle *handle, GError **error);
>       |          ^~~~~~~~~~~~~~~~~
<snip>

Same for me; see https://sourceforge.net/p/tuxpaint/bugs/278/ :)

I still need to work on it.


> src/tuxpaint.c:26894:15: warning: assignment discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers]
> 26894 |       ttffont = TTF_FontFaceFamilyName(getfonthandle(i)->ttf_font);
>       |               ^
> src/tuxpaint.c: In function 'load_embedded_data':

Casting to (char *); see if that helps.


> src/tuxpaint.c:27585:15: warning: pointer 'unc_buff' used after 'free' [-Wuse-after-free]
> 27585 |               free(unc_buff);
>       |               ^~~~~~~~~~~~~~
> src/tuxpaint.c:27581:15: note: call to 'free' here
> 27581 |               free(unc_buff);
>       |               ^~~~~~~~~~~~~~
> src/tuxpaint.c:27646:15: warning: pointer 'unc_buff' used after 'free' [-Wuse-after-free]
> 27646 |               free(unc_buff);
>       |               ^~~~~~~~~~~~~~
> src/tuxpaint.c:27642:15: note: call to 'free' here
> 27642 |               free(unc_buff);
>       |               ^~~~~~~~~~~~~~

I think I fixed it; see if that helps. :)


> src/fonts.c: In function 'TuxPaint_Font_OpenFont':
> src/fonts.c:342:16: warning: assignment discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers]
>   342 |     familyname = TTF_FontFaceFamilyName(ttf_font);      /* N.B.: I don't believe we're supposed to free() this... -bjk 2021.10.26 */
>       |                ^

[See above]


> src/fonts.c: In function 'getfonthandle':
> src/fonts.c:1357:20: warning: the comparison will always evaluate as 'true' for the address of 'filename' will never be NULL [-Waddress]
>  1357 |   if (fi->filename != NULL)
>       |                    ^~
> In file included from src/fonts.c:60:
> src/fonts.h:152:9: note: 'filename' declared here
>   152 |   char *filename[4];
>       |         ^~~~~~~~

Attempted a fix (was supposed to be indexing into that array).


> src/dirwalk.c: In function 'tp_ftw':
> src/dirwalk.c:436:2: warning: #warning Failed to see DT_UNKNOWN [-Wcpp]
>   436 | #warning Failed to see DT_UNKNOWN
>       |  ^~~~~~~

This is another case of us echoing a warning.  The comment here is:

  // Linux and BSD can often provide file type info w/o the stat() call

Then it tests for that existing, and does a 'switch/case' block,
testing the "d_type" of the file we see with 'readdir()'.

I don't really remember what's supposed to be going on in here
(or, in fact, if I had much to do with it!  20 years is a long time!),
so I'm going to ignore this one as well.  :shrug:

Please tell me if it throws fewer warnings now, and hopefully
also still builds and runs properly.

This goes for everyone!  (Mark, Pere, Luc :) )

Thanks!

-- 
-bill!
Sent from my computer