Mailing-List: contact dev-help@apr.apache.org; run by ezmlm
Precedence: bulk
Message-ID: <402389CE.50901@algroup.co.uk>
Date: Fri, 06 Feb 2004 12:34:22 +0000
From: Ben Laurie <ben@algroup.co.uk>
User-Agent: Mozilla Thunderbird 0.5b (Windows/20040204)
MIME-Version: 1.0
To: Joe Orton <jorton@redhat.com>
Cc: dev@apr.apache.org
Subject: Re: [PATCH] optional hook macros
References: <20040205171129.GA29555@manyfish.co.uk>
 <40236716.20705@algroup.co.uk> <20040206103107.GA13618@redhat.com>
In-Reply-To: <20040206103107.GA13618@redhat.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit

Joe Orton wrote:
> On Fri, Feb 06, 2004 at 10:06:14AM +0000, Ben Laurie wrote:
>>Surely the point is that the function and the function pointer actually 
>>_do_ have compatible types, but we hold the pointer in a variable that 
>>doesn't. That is, we cast it to an incompatible type for storage, then 
>>cast it back.
> 
> The issue is the call to apr_dynamic_fn_register, here's how I read it:
> 
> (((void (*)(const char *, APR_OPTIONAL_FN_TYPE(name) *)) 
>                &apr_dynamic_fn_register)(#name,name))
> 
> this takes a function pointer (&apr_dynamic_fn_register), then casts it
> to a different type (void (*)(const blah blah)), then calls the function
> using this type. That has undefined behaviour, since the type of
> apr_dynamic_fn_register is not compatible with the (void (*)(const blah
> blah) type.

I misunderstood, sorry.

Cheers,

Ben.

-- 
http://www.apache-ssl.org/ben.html       http://www.thebunker.net/

"There is no limit to what a man can do or how far he can go if he
doesn't mind who gets the credit." - Robert Woodruff