Re: [PATCH 1/2] Add support for setting disassembler-options in GDB for POWER, ARM and S390

[Alan Modra <amodra@gmail.com>]

On Tue, Jan 24, 2017 at 09:59:18AM -0600, Peter Bergner wrote:
> On 1/23/17 10:32 PM, Peter Bergner wrote:
> > On 1/23/17 9:36 PM, Alan Modra wrote:
> >>>> This appears to have lost translation of the help strings.
> > [snip]
> >> Yes, I meant the support for language translation.
> > 
> > Ok, I see now.  How about the following changes to to add _(...)
> > around the option descriptions for s390 and arm (ppc doesn't have any)?
> 
> For completeness, here is the full updated patch that keeps the language
> translation.

You're going to hate me, but this still doesn't keep the language
translations.  If you regenerate opcodes/po/opcodes.pot you'll see
that lines like

msgid "  esa         Disassemble in ESA architecture mode\n"

have disappeared without being replaced by anything.  I think you need
to wrap your new options[].description with N_() to let gettext know
they should be translated, like this:

static const options_t options[] =
{
  { "esa" ,       N_("Disassemble in ESA architecture mode") },
  { "zarch",      N_("Disassemble in z/Architecture mode") },
  { "insnlength", N_("Print unknown instructions according to "
		     "length from first two bits") }
};

The same thing needs to be done for regnames description in arm-dis.c.

I also see this arm-dis.c hunk
> @@ -6855,7 +6889,7 @@ the -M switch:\n"));
>      fprintf (stream, "  reg-names-%s %*c%s\n",
>  	     regnames[i].name,
>  	     (int)(14 - strlen (regnames[i].name)), ' ',
> -	     regnames[i].description);
> +	     _(regnames[i].description));
>  
>    fprintf (stream, "  force-thumb              Assume all insns are Thumb insns\n");
>    fprintf (stream, "  no-force-thumb           Examine preceding label to determine an insn's type\n\n");

showing missing _() for the last two lines.  Bonus points for fixing
them in a way that uses exactly the same translatated strings as you
have in disassembler_options_desc_arm.

The opcodes patch is OK with those changes.

-- 
Alan Modra
Australia Development Lab, IBM