[LONG MAIL]: The story of the tiny TAPE_GET_MEDIA_PARAMETERS searching the man in the moon

[Chris Faylor]

On Fri, Jun 15, 2001 at 06:10:08PM +0200, Corinna Vinschen wrote:
>On Fri, Jun 15, 2001 at 11:16:27AM -0400, Chris Faylor wrote:
>> On Fri, Jun 15, 2001 at 05:14:39PM +0200, Corinna Vinschen wrote:
>> >On Fri, Jun 15, 2001 at 11:00:27AM -0400, Chris Faylor wrote:
>> >> On Fri, Jun 15, 2001 at 12:39:07PM +0200, Corinna Vinschen wrote:
>> >> >At least the current native gcc-2.95.3 doesn't have it. Otherwise
>> >> >mt would have used the same wrong value of 28.
>> >> 
>> >> I use 2.95.3 to build the DLL.  I use the released gcc to build everything.
>> >
>> >Native or cross?
>> 
>> linux.
>
>So, when the result is different using the same winnt.h (which
>excludes that an explicit structure packing is different) and
>the result is different using the same compiler version on
>two different hosts, isn't it very likely that the structure
>packing algorithm in gcc >= 2.95.3 is somehow related to the host
>instead of to the target by mistake? Or do I overrate the
>phenomenon?
>
>We could workaround that of course by setting an explicit
>
>#define pack(push,8)
>...
>#define pack(pop)
>
>in winnt.h but that seems not right to me. The original WinNT.h
>from the Platform SDK doesn't contain that either.

I *don't know* what the problem is.  I would appreciate it if you
or someone else would track it down.

This doesn't rely on any special knowledge that only I have so I
think we can get insight from cygwin-developers.  I've Cc'ed them.

cgf