Re: portability issues with unicodeio

[Bruno Haible]

Kiyoshi KANAZAWA wrote:
> make check passed both on Solaris 11.3 & 11.4.

Oh, I see. So my unit test was not complete.

> > It is succeeds this test, then what is the difference between the
> > coreutils printf program and the test-unicodeio.c program? Both call
> > setlocale (LC_ALL, "").
> 
> I do not understand.
> What should I do next ?

I single-stepped both the coreutils printf program and the test-unicodeio.c
program, and found the issue.


2020-07-09  Bruno Haible  <bruno@clisp.org>

        unicodeio: Fix wrong result on Solaris 11.
        Reported by Kiyoshi Kanazawa <yoi_no_myoujou@yahoo.co.jp>
        via Akim Demaille <akim.demaille@gmail.com> in
        <https://lists.gnu.org/archive/html/bug-gnulib/2020-07/msg00036.html>.
        * lib/unicodeio.c (unicode_to_mb): Handle question mark fallback
        characters on Solaris.
        * tests/test-unicodeio.c (main): In the "C" locale, expect either the
        UTF-8 output or the specified fallback.

diff --git a/lib/unicodeio.c b/lib/unicodeio.c
index 8ae3308..81fe0dd 100644
--- a/lib/unicodeio.c
+++ b/lib/unicodeio.c
@@ -134,6 +134,10 @@ unicode_to_mb (unsigned int code,
 # if !defined _LIBICONV_VERSION && (defined sgi || defined __sgi)
           || (res > 0 && code != 0 && outptr - outbuf == 1 && *outbuf == '\0')
 # endif
+          /* Solaris 11 iconv() inserts a '?' if it cannot convert. */
+# if !defined _LIBICONV_VERSION && defined __sun
+          || (res > 0 && code != 0 && outptr - outbuf == 1 && *outbuf == '?')
+# endif
          )
         return failure (code, NULL, callback_arg);
 
diff --git a/tests/test-unicodeio.c b/tests/test-unicodeio.c
index be91a91..fa28e98 100644
--- a/tests/test-unicodeio.c
+++ b/tests/test-unicodeio.c
@@ -62,6 +62,7 @@ main (int argc, char *argv[])
       if (argc > 1)
         switch (argv[1][0])
           {
+          case '1': /* On some platforms, the "C" locale has UTF-8 encoding.  
*/
           case '2':
             ASSERT (strcmp (result, TEST_CODE_AS_UTF8) == 0);
             break;