better name for var_integer et.al.

better name for var_integer et.al.

[Andrew Cagney]

Hello,

GDB has the variable types:

  /* Unsigned Integer.  *VAR is an unsigned int.  The user can type 0
     to mean "unlimited", which is stored in *VAR as UINT_MAX.  */
  var_uinteger,

  /* Like var_uinteger but signed.  *VAR is an int.  The user can type 0
     to mean "unlimited", which is stored in *VAR as INT_MAX.  */
  var_integer,

  /* ZeroableInteger.  *VAR is an int.  Like Unsigned Integer except
     that zero really means zero.  */
  var_zinteger,

I think calling something "integer" or "unsigned integer" but then not 
allowing the value zero is just plain weird.

Would anyone have better names for the first two?

Andrew

Re: better name for var_integer et.al.

[Joel Brobecker]

>  /* Unsigned Integer.  *VAR is an unsigned int.  The user can type 0
>     to mean "unlimited", which is stored in *VAR as UINT_MAX.  */
>  var_uinteger,

Ada calls such numbers "Positive". var_positive might be a good name.

>  /* Like var_uinteger but signed.  *VAR is an int.  The user can type 0
>     to mean "unlimited", which is stored in *VAR as INT_MAX.  */
>  var_integer,

I can't see any use for this semantics, but maybe it's due to my limited
experience. I looked at the current code, and most if not all of them
where just misuses of this kind. Some of them are really booleans (so I
suspect var_zinteger would be better), or postive numbers (so
var_positive would be better).

I not useful, I would consider just removing it.

>  /* ZeroableInteger.  *VAR is an int.  Like Unsigned Integer except
>     that zero really means zero.  */
>  var_zinteger,

var_integer? (assuming we get rid of the non-zero signed integer)


-- 
Joel

Re: better name for var_integer et.al.

[Andrew Cagney]

>>>  /* Unsigned Integer.  *VAR is an unsigned int.  The user can type 0
>>>     to mean "unlimited", which is stored in *VAR as UINT_MAX.  */
>>>  var_uinteger,
> 
> 
> Ada calls such numbers "Positive". var_positive might be a good name.

Or var_ordinal_number (vs cardinal number)?

>>>  /* Like var_uinteger but signed.  *VAR is an int.  The user can type 0
>>>     to mean "unlimited", which is stored in *VAR as INT_MAX.  */
>>>  var_integer,

Well, the "set backtrace limit 100" bug comes from a comparison between 
signed (frame->limit == -1) vs unsigned (backtrace_limit == 100) 
comparison which is from a var_uinteger.

Using var_integer "fixes" it but lets a user enter -100.

> I can't see any use for this semantics, but maybe it's due to my limited
> experience. I looked at the current code, and most if not all of them
> where just misuses of this kind. Some of them are really booleans (so I
> suspect var_zinteger would be better), or postive numbers (so
> var_positive would be better).
> 
> I not useful, I would consider just removing it.
> 
> 
>>>  /* ZeroableInteger.  *VAR is an int.  Like Unsigned Integer except
>>>     that zero really means zero.  */
>>>  var_zinteger,
> 
> 
> var_integer? (assuming we get rid of the non-zero signed integer)

Andrew

Re: better name for var_integer et.al.

[Joel Brobecker]

> >Ada calls such numbers "Positive". var_positive might be a good name.
> 
> Or var_ordinal_number (vs cardinal number)?

Yew! (sorry).

I prefer var_positive. Or var_nonzero_positive.

> >>> /* Like var_uinteger but signed.  *VAR is an int.  The user can type 0
> >>>    to mean "unlimited", which is stored in *VAR as INT_MAX.  */
> >>> var_integer,
> 
> Well, the "set backtrace limit 100" bug comes from a comparison between 
> signed (frame->limit == -1) vs unsigned (backtrace_limit == 100) 
> comparison which is from a var_uinteger.
> 
> Using var_integer "fixes" it but lets a user enter -100.

What would it mean, at the semantic level if the user entered
such a value? Shouldn't this value always be positive, with a
zero value being infinity?

Anyway, if we need to keep this kind of variables, how about

var_nonzero_integer (in line with one of the proposals above).

> >>> /* ZeroableInteger.  *VAR is an int.  Like Unsigned Integer except
> >>>    that zero really means zero.  */
> >>> var_zinteger,
> >
> >
> >var_integer? (assuming we get rid of the non-zero signed integer)

And then this one could become var_integer   (an integer range contains
value zero by default, so no need to emphasize it with the 'z', IMHO).

-- 
Joel