Silence compiler warnings

[stefanrueger]

Fixes #1799

The %hd converts the int argument to short before printing, so I suspect #1799 is a matter of overzealous compiler warning.

[MCUdude]

Regarding compiler warnings, I'm getting the following one when building git main. No big deal, but I thought I'd mention it.

  LEX      lexer.c
/Library/Developer/CommandLineTools/usr/bin/make  all-recursive
Making all in .
  CC       avrdude-main.o
  CC       avrdude-whereami.o
  CC       avrdude-developer_opts.o
  CC       libavrdude_a-config_gram.o
  CC       libavrdude_a-lexer.o
lexer.c:2434:38: warning: comparison of integers of different signs: 'unsigned long' and 'int' [-Wsign-compare]
        if (((yy_n_chars) + number_to_move) > YY_CURRENT_BUFFER_LVALUE->yy_buf_size) {
             ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~  ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated.

[stefanrueger]

@MCUdude That's a known issue and we cannot do anything about it, as it's MacOS's flex that generates the code.

[dl8dtl]

Shouldn't we also remove the h and (unsigned short) cast here?

[stefanrueger]

No, that would change what happens. Right now printf's %hx will (unsigned short) the promoted int argument that itself already has been (unsigned short)-ed. We can remove either the h or the (unsigned short) cast of the (assuming at most int-wide) argument and the print will be the same. If we remove both then something else will be printed if the arg doesn't fit in an unsigned short.

[stefanrueger]

Here an example:

#include <stdio.h>
int main() {
  int i = 0x78000;
  printf("%04hx\n", (unsigned short) i);
  printf("%04x\n", (unsigned short) i);
  printf("%04hx\n", i);
  printf("%04x\n", i);
}

[mcuee]

MinGW build still has some warnings.
https://github.com/avrdudes/avrdude/actions/runs/8870239966/job/24351804351

[ 68%] Building C object src/CMakeFiles/libavrdude.dir/updi_link.c.obj
In file included from D:/a/avrdude/avrdude/src/updi_link.c:36:
D:/a/avrdude/avrdude/src/updi_link.c: In function 'updi_physical_send':
D:/a/avrdude/avrdude/src/updi_link.c:93:14: warning: format '%lu' expects argument of type 'long unsigned int', but argument 8 has type 'size_t' {aka 'long long unsigned int'} [-Wformat=]
   93 |   pmsg_debug("sending %lu bytes [", len);
      |              ^~~~~~~~~~~~~~~~~~~~~  ~~~
      |                                     |
      |                                     size_t {aka long long unsigned int}
D:/a/avrdude/avrdude/src/avrdude.h:91:138: note: in definition of macro 'pmsg_debug'
   91 | #define pmsg_debug(...)     avrdude_message2(stderr, __LINE__, __FILE__, __func__, MSG2_PROGNAME|MSG2_FLUSH|MSG2_LEFT_MARGIN, MSG_DEBUG, __VA_ARGS__)
      |                                                                                                                                          ^~~~~~~~~~~
D:/a/avrdude/avrdude/src/updi_link.c:93:25: note: format string is defined here
   93 |   pmsg_debug("sending %lu bytes [", len);
      |                       ~~^
      |                         |
      |                         long unsigned int
      |                       %llu
D:/a/avrdude/avrdude/src/updi_link.c: In function 'updi_physical_recv':
D:/a/avrdude/avrdude/src/updi_link.c:117:14: warning: format '%lu' expects argument of type 'long unsigned int', but argument 8 has type 'size_t' {aka 'long long unsigned int'} [-Wformat=]
  117 |   pmsg_debug("received %lu bytes [", len);
      |              ^~~~~~~~~~~~~~~~~~~~~~  ~~~
      |                                      |
      |                                      size_t {aka long long unsigned int}
D:/a/avrdude/avrdude/src/avrdude.h:[91](https://github.com/avrdudes/avrdude/actions/runs/8870239966/job/24351804351#step:5:92):138: note: in definition of macro 'pmsg_debug'
   91 | #define pmsg_debug(...)     avrdude_message2(stderr, __LINE__, __FILE__, __func__, MSG2_PROGNAME|MSG2_FLUSH|MSG2_LEFT_MARGIN, MSG_DEBUG, __VA_ARGS__)
      |                                                                                                                                          ^~~~~~~~~~~
D:/a/avrdude/avrdude/src/updi_link.c:117:26: note: format string is defined here
  117 |   pmsg_debug("received %lu bytes [", len);
      |                        ~~^
      |                          |
      |                          long unsigned int
      |                        %llu

[dl8dtl]

size_t {aka long long unsigned int}

size_t requires format letter z

[stefanrueger]

size_t requires format letter z

Correct; this was introduced with C99, but we have in the past tried to be even compatible with pre-1999 C-libraries (#1608) so probably better to cast len to (unsigned long) in this case. Just pushed a commit for this

[dl8dtl]

I wouldn't do so unless MSVC is still that far behind to require it.
We're requiring a lot of C99 already elsewhere anyway.

[stefanrueger]

The C99 discussion was quite recent. Full details at #1594. I don't recall the details, but think someone built AVRDUDE in Windows 7 (that couldn't be upgraded) and the only problem they faced was %z[idux] conversion. Apparently, the rest of the code was fine (at least what they needed) and it isn't much effort to do without %z[idux].

[dl8dtl]

Quite sad, it's only 15 years that have gone since …

[mcuee]

@stefanrueger and @dl8dtl

My suggestion is to give up on that paticular configuration -- mingw32 cross compiler toolchain under Windows 7.

If you read my comments in #1594, there is no issue using MSVC build and normal MSYS2 mingw32/64 build.

[mcuee]

I wouldn't do so unless MSVC is still that far behind to require it. We're requiring a lot of C99 already elsewhere anyway.

MSVC is not the problem here. So one less worry.

[mcuee]

@dl8dtl and @stefanrueger and @MCUdude

Quite sad, it's only 15 years that have gone since …

Let's face it, Windows 7 was released on 22-Sept-2009 and Microsoft stopped the extended support on 10-Jan-2023 (main line support ended on 14-Jan-2020). I think it is perfectly fine that we drop the support of Windows 7 altogether.

I do not have Windows 7 machine and I am not so sure if anyone of use still wants to carry out test under Windows 7.

Same for Windows 8/8.1 -- both reached EoL status as well.
https://learn.microsoft.com/en-us/lifecycle/faq/windows
https://support.microsoft.com/en-us/windows/windows-8-1-support-ended-on-january-10-2023-3cfd4cde-f611-496a-8057-923fba401e93

Windows 8.1 reached the end of Mainstream Support on January 9, 2018, and reached the end of Extended Support on January 10, 2023.

[dl8dtl]

You're right about Win7. But then, I think it's fine to use the 'z' modifier, rather than casting stuff around.

[mcuee]

You're right about Win7. But then, I think it's fine to use the 'z' modifier, rather than casting stuff around.

No issue with that. I just want to establish that we will not officially support Windows 7/8/8.1 from now on. It may still work well. However, if there are issues, we may rely on the users to provide non-intrusive patches if they really care about Windows 7/8/8.1.

BTW, MSYS2 has dropped 32bit Windows support of libusb/libusb-compat/libftdi, so we have already dropped 32bit mingw support of avrdude as well. Take note Windows 10/11 only supports x86_64 and not 32bit x86.

[mcuee]

@stefanrueger and @dl8dtl

Not sure if we can fix MSYS2 clang64 compiler warnings as well.
https://github.com/avrdudes/avrdude/actions/runs/8935908735/job/24545265227

D:/a/avrdude/avrdude/src/ser_win32.c:436:3: warning: comparison of integers of different signs: 'SOCKET' (aka 'unsigned long long') and 'const int' [-Wsign-compare]
  436 |                 FD_SET(fd->ifd, &rfds);
      |                 ^~~~~~~~~~~~~~~~~~~~~~
D:/a/_temp/msys64/clang64/include/psdk_inc/_fd_types.h:77:40: note: expanded from macro 'FD_SET'
   77 |                 if (((fd_set *)(set))->fd_array[__i] == (fd)) {         \
      |                     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^   ~~
D:/a/avrdude/avrdude/src/ser_win32.c:[55](https://github.com/avrdudes/avrdude/actions/runs/8935908735/job/24545265227#step:5:56)7:3: warning: comparison of integers of different signs: 'SOCKET' (aka 'unsigned long long') and 'const int' [-Wsign-compare]
  557 |                 FD_SET(fd->ifd, &rfds);
      |                 ^~~~~~~~~~~~~~~~~~~~~~
D:/a/_temp/msys64/clang64/include/psdk_inc/_fd_types.h:77:40: note: expanded from macro 'FD_SET'
   77 |                 if (((fd_set *)(set))->fd_array[__i] == (fd)) {         \
      |                     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^   ~~
2 warnings generated.

[stefanrueger]

Not sure if we can fix MSYS2 clang64 compiler warnings as well.

See here