SRFI-4's `scm_take_TAGvector ()'

SRFI-4's `scm_take_TAGvector ()'

[Ludovic Courtès]

Hi,

While the name of the `scm_take_TAGvector ()' function is pretty good at
suggesting that it will take control over the data pointed to by the
first argument, I find the prototype of these functions quite
misleading.  IMO, instead of:

  SCM scm_take_u8vector (const scm_t_uint8 *data, size_t len);

this should be:

  SCM scm_take_u8vector (scm_t_uint8 *data, size_t len);

Removing the `const' qualifier provides additional semantic information
to the compiler that could help detect misuses of these functions.  For
example, a warning would be issued for a call like:

  scm_take_u8vector ("hello", 6);

What do you think?

Thanks,
Ludovic.


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: SRFI-4's `scm_take_TAGvector ()'

[Ken Raeburn]

On Aug 17, 2005, at 10:16, Ludovic Courtès wrote:
> While the name of the `scm_take_TAGvector ()' function is pretty  
> good at
> suggesting that it will take control over the data pointed to by the
> first argument, I find the prototype of these functions quite
> misleading.  IMO, instead of:
>
>   SCM scm_take_u8vector (const scm_t_uint8 *data, size_t len);
>
> this should be:
>
>   SCM scm_take_u8vector (scm_t_uint8 *data, size_t len);
>
> Removing the `const' qualifier provides additional semantic  
> information
> to the compiler that could help detect misuses of these functions.   
> For
> example, a warning would be issued for a call like:
>
>   scm_take_u8vector ("hello", 6);

... except for the little detail that ISO C says a string literal has  
type "char [N]", not "const char [N]".  So, probably no warning in  
this case.

_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: SRFI-4's `scm_take_TAGvector ()'

[Ludovic Courtès]

Ken Raeburn <raeburn@gnu.org> writes:

> ... except for the little detail that ISO C says a string literal has  
> type "char [N]", not "const char [N]".  So, probably no warning in  
> this case.

Indeed!  One never stops learning C...  ;-)

Still removing `const' makes sense in the general case, don't you think
so?

Thanks,
Ludovic.


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: SRFI-4's `scm_take_TAGvector ()'

[Marius Vollmer]

ludovic.courtes@laas.fr (Ludovic Courtès) writes:

> Hi,
>
> While the name of the `scm_take_TAGvector ()' function is pretty good at
> suggesting that it will take control over the data pointed to by the
> first argument, I find the prototype of these functions quite
> misleading.  IMO, instead of:
>
>   SCM scm_take_u8vector (const scm_t_uint8 *data, size_t len);
>
> this should be:
>
>   SCM scm_take_u8vector (scm_t_uint8 *data, size_t len);
>
> What do you think?

I agree.  I made that change.

(Slowly working towards your other patches now... hang in there...)

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user