Neil Brown wrote:
> On Friday August 3, chuck.lever@oracle.com wrote:
>> Previous NLS changes missed a spot or two.  This patch tries to get
>> most of them, but probably misses a few more.
>>
>> In errors.c:mount_errors() I've removed a period at the end of the error
>> messages; this is consistent with other error messages I've examined.
>>
>> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> ....
>> --- a/utils/mount/nfs4mount.c
>> +++ b/utils/mount/nfs4mount.c
>> @@ -321,8 +321,8 @@ int nfs4mount(const char *spec, const char *node, int flags,
>>  			else if (!strcmp(opt, "sharecache"))
>>  				unshared = !val;
>>  			else if (!sloppy) {
>> -				printf(_("unknown nfs mount option: "
>> -					 "%s%s\n"), val ? "" : "no", opt);
>> +				printf(_("unknown nfs mount option: %s%s\n"),
>> +						val ? "" : _("no"), opt);
>>  				goto fail;
> 
> I removed the _() around the "no".  Translating the 'no' without
> translating the option doesn't make any sense to me.
> You would run
>   mount -o nocoffee .....
> and get an error
>    Option inconnue de mount du nfs: noncoffee
> 
> (or something - my French non-existent, I just played with
> google-translate).  Which is wrong.  'noncoffee' isn't the bad
> option.  'nocoffee' is.

OK.

>> --- a/utils/mount/nfsumount.c
>> +++ b/utils/mount/nfsumount.c
>> @@ -67,7 +67,7 @@ static int del_mtab(const char *spec, const char *node)
>>                  res = umount2 (node, MNT_FORCE);
>>                  if (res == -1) {
>>                          int errsv = errno;
>> -                        perror("umount2");
>> +                        perror(_("umount2"));
>>                          errno = errsv;
>>                          if (errno == ENOSYS) {
>>                                  if (verbose)
>> @@ -93,7 +93,7 @@ static int del_mtab(const char *spec, const char *node)
>>                                  update_mtab(node, &remnt);
>>                          return 0;
>>                  } else if (errno != EBUSY) {    /* hmm ... */
>> -                        perror("remount");
>> +                        perror(_("remount"));
>>                          nfs_error(_("%s: could not remount %s read-only"),
>>                                  	progname, spec);
>>                  }
> 
> And I'm not at all convinced about the _() in perror - In the first
> case it is a syscall name.  How would you translate that?  The second
> is possibly less doubtful, but it would probably be best to discard
> the perror calls and pass strerror(errno) to nfs_error with an
> appropriate format - What do you think?

Adding _() for perror is consistent with the other perror calls already 
in the code.  I suppose if those are objectionable, you should remove 
them all.