Bardur Arantsson wrote:
> Bardur Arantsson wrote:
>> Patrice Dumas wrote:
> [--snip--]
>
> (I'm going to split my reply up because some of this could go on for a
> while... I'll probably split the reply up over a couple of days as well.)
>
>>>> The macro KEY_METAINFO_SIZE also has issues on my AMD64:
>>>>
>>>> #define KEY_METAINFO_SIZE(k) ((unsigned int)&(k->recordSize) -
>>>> unsigned int)k)
>>>>
>>>> is incorrect when the result is used as a "size_t" in various
>>>> locations.
>>>> The correct definition is
>>>>
>>>> #define KEY_METAINFO_SIZE(k) ((size_t)&(k->recordSize) - (size_t)k)
>>> Since recordSize is size_t in _Key, wouldn't it be even better to
>>> use
>>> #define KEY_METAINFO_SIZE(k) ((k->recordSize) - (size_t)k)
>>>
>>
>> Yup, I hadn't checked the definition of recordSize.
>
> Reinvestigating this, the proper definition is as follows:
>
> #define KEY_METAINFO_SIZE(k) ((size_t) (((char *) (&k->recordSize))
> - (char *) k)))
>
Gah. Apparently the header file change didn't cause make to rebuild all
the affected .c files. This still results in a warning from gcc about
(size_t) being smaller than the pointer type.
The old code is subtly incorrect, so the patch should still be applied.
I don't care enough to investigate further how the warning can be removed.
--
Bardur Arantsson
<[EMAIL PROTECTED]>
- Let your anger be like a monkey in a piñata.
The Old Master / Kung Pow
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys -- and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Registry-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/registry-list
