[lucy-dev] Const-ness and "steal" String constructors

[Marvin Humphrey <ma...@rectangular.com>]

On Sat, Sep 14, 2013 at 12:29 PM,  <nw...@apache.org> wrote:
> Const 'ptr' instance variable of String

> Project: http://git-wip-us.apache.org/repos/asf/lucy/repo
> Commit: http://git-wip-us.apache.org/repos/asf/lucy/commit/fc363f05
> Tree: http://git-wip-us.apache.org/repos/asf/lucy/tree/fc363f05
> Diff: http://git-wip-us.apache.org/repos/asf/lucy/diff/fc363f05

>      /** Return a new String which holds a copy of the passed-in string.
>       * Check for UTF-8 validity.
> @@ -48,19 +48,19 @@ class Clownfish::String cnick Str
>       * passed-in string.  Check validity of supplied UTF-8.
>       */
>      inert incremented String*
> -    new_steal_str(char *ptr, size_t size);
> +    new_steal_str(const char *ptr, size_t size);
>
>      /** Return a pointer to a new String which assumes ownership of the
>       * passed-in string.  Do not check validity of supplied UTF-8.
>       */
>      inert incremented String*
> -    new_steal_from_trusted_str(char *ptr, size_t size);
> +    new_steal_from_trusted_str(const char *ptr, size_t size);
>
>      /** Initialize the String using the passed-in string.  Do not check
>       * validity of supplied UTF-8.
>       */
>      public inert String*
> -    init_steal_trusted_str(decremented String *self, char *ptr, size_t size);
> +    init_steal_trusted_str(String *self, const char *ptr, size_t size);
>
>      /** Return a String which holds a single character.

It seems to me that the "steal" constructors should take `char*` instead of
`const char*`.  We have to cast away constness in the destructor when we
invoke "free" on the stolen strings, right?  The potential exists to create a
situation where we free() non-heap memory and the compiler doesn't issue a
warning.

Marvin Humphrey

Re: [lucy-dev] Const-ness and "steal" String constructors

[Nick Wellnhofer <we...@aevum.de>]

On 10/10/2013 05:11, Marvin Humphrey wrote:
> On Sat, Sep 14, 2013 at 12:29 PM,  <nw...@apache.org> wrote:
>>       inert incremented String*
>> -    new_steal_str(char *ptr, size_t size);
>> +    new_steal_str(const char *ptr, size_t size);
>
> It seems to me that the "steal" constructors should take `char*` instead of
> `const char*`.  We have to cast away constness in the destructor when we
> invoke "free" on the stolen strings, right?  The potential exists to create a
> situation where we free() non-heap memory and the compiler doesn't issue a
> warning.

You're right. Good catch.

Nick