Compile S_emulate_langinfo() everywhere where needed

[khwilliamson]

Prior to this series of commits, this function was not getting compiled in some Configurations where it was needed, and not getting called always when needed.

[jkeenan]

Point of information for a non-C programmer: Is there a difference between this syntax:

# ifdef SOMETHING

and this syntax:

# if defined(SOMETHING)

Why do you use one or the other?

[Tux]

#ifdef can only deal with one singe symbol
#if defined(FOO) is equal to #ifdef FOO, but when testing on more than one single symbol, like in this code, the second syntax is required
#if defined(FOO) || defined(BAR) does not have an #ifdef variant

[mauke]

#ifdef FOO is subsumed by the form #if defined FOO (or equivalently, #if defined ( FOO )). As for why both forms exist, I believe #ifdef (and #ifndef) came first historically and defined was added later when people needed more complex conditionals.

Choosing between them is mostly a matter of taste (personally I don't like the extra parentheses in the defined (FOO) form, for example). Some purists argue that you should limit yourself to #ifdef because anything more complex than that is a sign that you're abusing the preprocessor for things it was never meant to do (and you should presumably move complex configuration logic to a separate stage that emits simple macro definitions). In perl, that train has sailed.

[jkeenan]

As reported in #21909 (comment), the 3 commits in this pull request resolve the build-time failure reported in #21909 but generate 3 new build-time warnings. (The commit IDs below are from my local research branch.)

$ report-build-warnings 7d786c0271.linux.unthreaded.no-locale-messages.maketp.output.txt.gz 
File:  7d786c0271.linux.unthreaded.no-locale-messages.maketp.output.txt.gz

  Wunused-variable                           3

[make-output] 2020 $ parse-build-warnings 7d786c0271.linux.unthreaded.no-locale-messages.maketp.output.txt.gz 
File:  7d786c0271.linux.unthreaded.no-locale-messages.maketp.output.txt.gz

[
  {
    char   => 9,
    group  => "Wunused-variable",
    line   => 6504,
    source => "locale.c",
    text   => "unused variable \xE2\x80\x98localeconv_klen\xE2\x80\x99",
  },
  {
    char   => 18,
    group  => "Wunused-variable",
    line   => 6503,
    source => "locale.c",
    text   => "unused variable \xE2\x80\x98localeconv_key\xE2\x80\x99",
  },
  {
    char   => 28,
    group  => "Wunused-variable",
    line   => 6502,
    source => "locale.c",
    text   => "unused variable \xE2\x80\x98cat_index\xE2\x80\x99",
  },
]

From the output of make test_prep:

cc -c -DPERL_CORE -fwrapv -fno-strict-aliasing -pipe -fstack-protector-strong -I/usr/local/include -
D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -std=c99 -O2 -Wall -Werror=pointer-arith -Werror=vla -Wext
ra -Wno-long-long -Wno-declaration-after-statement -Wc++-compat -Wwrite-strings locale.c
locale.c: In function ‘S_emulate_langinfo’:
locale.c:6504:9: warning: unused variable ‘localeconv_klen’ [-Wunused-variable]
 6504 |     I32 localeconv_klen;
      |         ^~~~~~~~~~~~~~~
locale.c:6503:18: warning: unused variable ‘localeconv_key’ [-Wunused-variable]
 6503 |     const char * localeconv_key;
      |                  ^~~~~~~~~~~~~~
locale.c:6502:28: warning: unused variable ‘cat_index’ [-Wunused-variable]
 6502 |     locale_category_index  cat_index;
      |                            ^~~~~~~~~

So we should eliminate these new build-time warnings before merging this branch to blead. Once the warnings are gone, the p.r. should be reviewed to see if it meets its own objectives and if the C-level code is correct.

[khwilliamson]

#21892 eliminates these warnings, in conjunction with this PR

[jkeenan]

#21892 eliminates these warnings, in conjunction with this PR

I continued with the QA procedure I described in #21909 (comment). I started with the investigative branch I described in that ticket and merged #21892 into it. I then configured and built twice, first with my regular (Linux) configuration switches, then adding -Accflags=-DNO_LOCALE_MESSAGES. In both cases, make test_prep was successful and emitted no warnings. I also ran the second case through make test_harness and got PASS.

Since we've been juggling quite a few locale-related pull requests lately, I recommend we get final review on this p.r. and #21892 (i.e., review by someone other than me), merge them into blead, and only then look to review additional p.r.s.