Re: [bug-GNUnet] [bug-gnunet] Spanish translation

[Miguel Ángel]

Hi Christian.

I did not join to bug-GNUnet (now I am) so I did not receive the last
reply message, but I have read it in the archive. Also I haven't
received the TP confirmation yet, as soon as I get it, I will notify
here.

The string "`%' must be followed by HEX number" can be found at:
./gnunet/src/fs/fs_uri.c:220:        *emsg = GNUNET_strdup (_("`%' must
be followed by HEX number"));

I have taking a look over the code and it is returned by the local
function percent_decode_keyword, that it is only used in the local
function uri_ksk_parse that is used at the function GNUNET_FS_uri_parse.
Every use of this return value is to a %s in a format string and (like
probably any other error message) it is not used as the format. There
may be a memory leak that I could look carefully at fs-api.c
  if (NULL != uris)
  {
    uc->ksk_uri = GNUNET_FS_uri_parse (uris, &emsg);
    GNUNET_free (uris);
    if (NULL == uc->ksk_uri)
    {
      GNUNET_break (0);
      goto cleanup;
    }
  }
Because emsg is also used in the cleanup here
cleanup:
  GNUNET_free_non_null (uc->filename);
  if ((NULL != rh) && (GNUNET_OK != GNUNET_BIO_read_close (rh, &emsg)))

Maybe I am loosing time because that function already frees the memory
or It is not actually being called, you may know better than me. I saw
it looking for the emsg printing, but I also could test it.

Resuming, that string is not a c-format string. But when gnunet.pot is
generated by make it contains:
#: src/fs/fs_uri.c:220
#, c-format
msgid "`%' must be followed by HEX number"
msgstr ""

That gives always an error when you execute msgfmt except when copying
the same string. It seems that gettext sees %' format as valid and
searches it into the msgstr. Here
(http://www.gnu.org/software/gettext/manual/html_node/c_002dformat-Flag.html) 
you can find information about that, the solution is to place around a comment 
with xgettext:no-c-format.

Happy hacking! :)
Miguel