Re: gas-2.20.90.pot errors

[Nick Clifton]

Hi Jorma,

  I have checked in a patch to fix the typos that you reported.  There 
were a couple of other questions though:

> #: config/tc-arm.c:2401
> #, c-format
> msgid "ignoring attempt to undefine built-in register '%s'"
>
> I do not understand this one. Should it be redefine, not undefine ?

Actually no, "undefine" is correct.  The message occurs when the user 
has something like this in their assembler source code:

  .unreq r1

The ".unreq" pseudo-op is used to remove a register name alias that was 
created by an earlier ".req" pseudo-op.  But if the user tries to use it 
to undefine one of the standard register names then they receive the 
warning message shown above.  Here is an example of the correct 
operation of the .unreq pseudo-op:

     arg1 .req r4
     arg2 .req r5

  foo:
     add   arg2, arg1
     rts

     .unreq arg1
     .unreq arg2

So arg1 and arg are aliases for registers 4 and 5, and they are being 
used to make the code more readable.  But they are undefined at the end 
of the definition of function foo, so that they will not affect code 
later on in the source file.

Perhaps a more understandable warning message would be:

  "ignoring attempt to use .unreq on a fixed register name"

What do you think ?



> #: config/tc-ia64.c:6989
[...]
> "  -xauto\t\t  automagically remove dependency violations (default)\n"
[...]
> Should it be automatically here, not automagically (in -xauto) ?

Yes and no.  "Automagically" is one of those computer geek puns.  It 
means "automatically and in a somewhat magical way".  It implies that 
not only will the feature happen without any user intervention, but also 
that it will do what the user wants without them even understanding how 
it works.

When translating, treating "automagically" as if it were "automatically" 
is perfectly acceptable.

Cheers
  Nick