From: Mats Kindahl
Date: August 27 2010 7:03am
Subject: Re: [STYLE] use true/false in C++, not TRUE/FALSE
List-Archive: http://lists.mysql.com/internals/38035
Message-Id: <4C77632A.1080608@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit

Hi Guilhem,

I support both proposals.

There is no problem with the standard conversions regarding C:s idea of
"true" to C++:s idea of "true". It was designed explicitly for this.

Just my few cents,
Mats Kindahl

On 08/26/2010 03:16 PM, Guilhem Bichot wrote:
> Hello,
>
> http://forge.mysql.com/wiki/MySQL_Internals_Coding_Guidelines
> says
>
> "Use TRUE and FALSE instead of true and false in C++ code. This makes
> the code more readable and makes it easier to use it later in a C
> library, if needed."
>
> Proposal 1
> ==========
> Use true/false, not TRUE/FALSE, for storing in into bool objects, or
> into bool function parameters. Do so for new code. Change existing
> code only when there's a deeper reason.
>
> Rationale
> =========
> I disagree with "more readable". I find that those capital letters are
> so eye-catching that they attract too much the reader's attention, for
> example when I read those lines (from 5.1) I pay more attention to
> TRUE that to "ref", which is bad:
>
>         if ((*group->item)->walk(&Item::find_item_processor, TRUE,
>                                  (uchar *) ref))
>
>     having->split_sum_func2(thd, ref_pointer_array, all_fields,
>                             &having, TRUE);
>
> Also, I find it strange: "bool" can take "true" or "false" by design,
> why are we forced to use something else?
>
> Proposal 2
> ==========
>
> Extend proposal 1 to using true/false for storing into my_bool C
> function parameters, and to using bool for storing the my_bool result
> of C functions.
>
> Rationale
> =========
> TRUE/FALSE is needed for my_bool (which is char) in C code.
> As for calls to C functions from C++, if the function takes a my_bool
> parameter, passing true/false is ok, as there would be implicit
> conversion from bool to my_bool, true becomes 1 (==TRUE), false
> becomes 0 (==FALSE). Storing a my_bool result from the C function into
> a bool would also be correct (any non-zero value would become true).
> Thus it sounds possible to completely eliminate TRUE/FALSE from C++ code.
>