Re: meson: problems building under msys2/mingw-w64 native

[Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>]

On 26/08/2020 16:34, Paolo Bonzini wrote:

> On Wed, Aug 26, 2020 at 12:15 AM Mark Cave-Ayland
> <mark.cave-ayland@ilande.co.uk> wrote:
>> ...
>> -rw-r--r-- 1 Mark None    39424 Aug 25 20:43 vgabios-stdvga.bin
>> -rw-r--r-- 1 Mark None    39424 Aug 25 20:43 vgabios-virtio.bin
>> -rw-r--r-- 1 Mark None    39424 Aug 25 20:43 vgabios-vmware.bin
>>
>>
>> I would expect the majority of these files to be in share/ or bin/ subdirectories as
>> they are in Linux.
> 
> This has always been different between Windows and Linux:
> 
> if test "$mingw32" = "yes" ; then
>     mandir="$prefix"
>     datadir="$prefix"
>     qemu_docdir="$prefix"
>     bindir="$prefix"
>     sysconfdir="$prefix"
>     local_statedir=
> else
>     mandir="${mandir:-$prefix/share/man}"
>     datadir="${datadir:-$prefix/share}"
>     qemu_docdir="${qemu_docdir:-$prefix/share/doc/qemu}"
>     bindir="${bindir:-$prefix/bin}"
>     sysconfdir="${sysconfdir:-$prefix/etc}"
>     local_statedir="${local_statedir:-$prefix/var}"
> fi
> 
> See also how os_find_datadir() is different between Windows and POSIX.
> I plan to fix this mess (and incidentally the pre_prefix hack) in 5.2
> by making the installation relocatable.

Ah so a long standing bug and not a regression - I can still test the builds so happy
to leave this for now and let you follow up with the relocation work later.


ATB,

Mark.