Thu Feb 02 11:55:59 2012: Request 74578 was acted upon.
Transaction: Correspondence added by rur...@cpanel.net
Queue: Win32-API
Subject: Re: [rt.cpan.org #74578] Win32::API::Struct not aligned on 64 bit
Broken in: 0.64
Severity: Normal
Owner: Nobody
Requestors: do...@cpan.org
Status: open
Ticket <URL: https://rt.cpan.org/Ticket/Display.html?id=74578 >
Am 01.02.2012 um 11:00 schrieb Cosimo Streppone:
> This is RT#74578 about Win32::API structs packing/unpacking.
> Can you help us?
>
> ------- Forwarded message -------
> From: "Douglas Wilson via RT" <bug-win32-...@rt.cpan.org>
> Subject: [rt.cpan.org #74578] Win32::API::Struct not aligned on 64 bit
> Date: Wed, 01 Feb 2012 17:27:37 +0100
>
> Queue: Win32-API
> Ticket <URL: https://rt.cpan.org/Ticket/Display.html?id=74578 >
>
> On Wed Feb 01 02:15:26 2012, COSIMO wrote:
>> Hi Doug, thanks for filing this.
>>
>> What do to about it?
>> Not sure, other than CC'ing libwin32@perl.org as you did.
>
> Cool, I didn't know RT cc'd the mailing list (maybe someone else will
> know what to do). What I've done is hack Struct.pm getPack() and
> getUnpack():
>
> in getPack():
> $type_align = (($packed_size + $type_size) % $type_size);
> # $packing .= "x" x $type_align . $type;
> $packing .= $type;
> $packed_size += $type_size;
> # $packed_size += $type_size + $type_align;
> }
>
> in getUnpack():
> $type_align = (($packed_size + $type_size) % $type_size);
> # $packing .= "x" x $type_align . $type;
> $packing .= $type;
> # $packed_size += $type_size + $type_align;
> $packed_size += $type_size;
>
> push(@items, $name);
> }
>
I'm pretty sure deleting type_align on 64bit is wrong. There must be an error
in the
size calculation for 8 byte sizes, which will result in the needed type_align=0.
I have got no win64 bit vm (yet), but I will try to study it today.
> Perhaps if this change were wrapped in is_64bit() type logic, then it
> would work everywhere? I don't know enough about this to really know for
> sure, and I'm not even sure if I haven't broken something else. And if
> this change is headed in the right direction, there should probably be
> some changes to the align() and sizeof() methods.
Reini
